diff --git a/crates/openfang-channels/src/line.rs b/crates/openfang-channels/src/line.rs
index b20294afc..ab5211f5a 100644
--- a/crates/openfang-channels/src/line.rs
+++ b/crates/openfang-channels/src/line.rs
@@ -108,7 +108,7 @@ impl LineAdapter {
             diff |= a ^ b;
         }
         if diff != 0 {
-            let computed = base64::engine::general_purpose::STANDARD.encode(&result);
+            let computed = base64::engine::general_purpose::STANDARD.encode(result);
             // Log first/last 4 chars of each signature for debugging without leaking full HMAC
             let comp_redacted = format!(
                 "{}...{}",
diff --git a/crates/openfang-cli/src/launcher.rs b/crates/openfang-cli/src/launcher.rs
index 18a8f1236..178d94d07 100644
--- a/crates/openfang-cli/src/launcher.rs
+++ b/crates/openfang-cli/src/launcher.rs
@@ -20,6 +20,7 @@ const PROVIDER_ENV_VARS: &[(&str, &str)] = &[
     ("ANTHROPIC_API_KEY", "Anthropic"),
     ("OPENAI_API_KEY", "OpenAI"),
     ("DEEPSEEK_API_KEY", "DeepSeek"),
+    ("VOLCENGINE_API_KEY", "Volcano Engine"),
     ("GEMINI_API_KEY", "Gemini"),
     ("GOOGLE_API_KEY", "Gemini"),
     ("GROQ_API_KEY", "Groq"),
@@ -31,7 +32,7 @@ const PROVIDER_ENV_VARS: &[(&str, &str)] = &[
 
 fn detect_provider() -> Option<(&'static str, &'static str)> {
     for &(var, name) in PROVIDER_ENV_VARS {
-        if std::env::var(var).is_ok() {
+        if std::env::var(var).ok().filter(|v| !v.is_empty()).is_some() {
             return Some((name, var));
         }
     }
diff --git a/crates/openfang-cli/src/main.rs b/crates/openfang-cli/src/main.rs
index 104286a8e..15d275ece 100644
--- a/crates/openfang-cli/src/main.rs
+++ b/crates/openfang-cli/src/main.rs
@@ -1420,6 +1420,18 @@ fn provider_list() -> Vec<(&'static str, &'static str, &'static str, &'static st
         ("groq", "GROQ_API_KEY", "llama-3.3-70b-versatile", "Groq"),
         ("gemini", "GEMINI_API_KEY", "gemini-2.5-flash", "Gemini"),
         ("deepseek", "DEEPSEEK_API_KEY", "deepseek-chat", "DeepSeek"),
+        (
+            "volcengine",
+            "VOLCENGINE_API_KEY",
+            "doubao-seed-1-6-251015",
+            "Volcano Engine",
+        ),
+        (
+            "volcengine_coding",
+            "VOLCENGINE_API_KEY",
+            "ark-code-latest",
+            "Volcano Engine Coding Plan",
+        ),
         (
             "anthropic",
             "ANTHROPIC_API_KEY",
@@ -4541,6 +4553,7 @@ fn provider_to_env_var(provider: &str) -> String {
         "perplexity" => "PERPLEXITY_API_KEY".to_string(),
         "cohere" => "COHERE_API_KEY".to_string(),
         "xai" => "XAI_API_KEY".to_string(),
+        "volcengine" | "doubao" | "volcengine_coding" => "VOLCENGINE_API_KEY".to_string(),
         "brave" => "BRAVE_API_KEY".to_string(),
         "tavily" => "TAVILY_API_KEY".to_string(),
         other => format!("{}_API_KEY", other.to_uppercase()),
@@ -4592,6 +4605,13 @@ pub(crate) fn test_api_key(provider: &str, env_var: &str) -> bool {
             .get("https://openrouter.ai/api/v1/models")
             .bearer_auth(&key)
             .send(),
+        "volcengine" | "doubao" => {
+            let base = openfang_types::model_catalog::VOLCENGINE_BASE_URL.trim_end_matches('/');
+            client.get(format!("{base}/models")).bearer_auth(&key).send()
+        }
+        // The Ark Coding endpoint (/api/coding/v3) does not expose a standard
+        // OpenAI-compatible /models list — skip probing and assume the key is valid.
+        "volcengine_coding" => return true,
         _ => return true, // unknown provider — skip test
     };
 
diff --git a/crates/openfang-cli/src/tui/screens/init_wizard.rs b/crates/openfang-cli/src/tui/screens/init_wizard.rs
index 279b1e6cf..e7ca8c456 100644
--- a/crates/openfang-cli/src/tui/screens/init_wizard.rs
+++ b/crates/openfang-cli/src/tui/screens/init_wizard.rs
@@ -68,6 +68,22 @@ const PROVIDERS: &[ProviderInfo] = &[
         needs_key: true,
         hint: "",
     },
+    ProviderInfo {
+        name: "volcengine",
+        display: "Volcano Engine",
+        env_var: "VOLCENGINE_API_KEY",
+        default_model: "doubao-seed-1-6-251015",
+        needs_key: true,
+        hint: "ByteDance Ark platform; cn-beijing; override base_url for other regions",
+    },
+    ProviderInfo {
+        name: "volcengine_coding",
+        display: "Volcano Engine (Coding Plan)",
+        env_var: "VOLCENGINE_API_KEY",
+        default_model: "ark-code-latest",
+        needs_key: true,
+        hint: "Shares VOLCENGINE_API_KEY with Volcano Engine standard plan. Uses Ark Coding endpoint.",
+    },
     ProviderInfo {
         name: "openrouter",
         display: "OpenRouter",
@@ -143,7 +159,7 @@ const PROVIDERS: &[ProviderInfo] = &[
     ProviderInfo {
         name: "qwen",
         display: "Qwen (Alibaba)",
-        env_var: "QWEN_API_KEY",
+        env_var: "DASHSCOPE_API_KEY",
         default_model: "qwen-plus",
         needs_key: true,
         hint: "",
@@ -151,7 +167,7 @@ const PROVIDERS: &[ProviderInfo] = &[
     ProviderInfo {
         name: "huggingface",
         display: "Hugging Face",
-        env_var: "HUGGINGFACE_API_KEY",
+        env_var: "HF_API_KEY",
         default_model: "meta-llama/Llama-3.3-70B-Instruct",
         needs_key: true,
         hint: "",
@@ -167,7 +183,7 @@ const PROVIDERS: &[ProviderInfo] = &[
     ProviderInfo {
         name: "replicate",
         display: "Replicate",
-        env_var: "REPLICATE_API_KEY",
+        env_var: "REPLICATE_API_TOKEN",
         default_model: "meta/meta-llama-3-70b-instruct",
         needs_key: true,
         hint: "",
diff --git a/crates/openfang-cli/src/tui/screens/welcome.rs b/crates/openfang-cli/src/tui/screens/welcome.rs
index 768a51ca4..96f94d115 100644
--- a/crates/openfang-cli/src/tui/screens/welcome.rs
+++ b/crates/openfang-cli/src/tui/screens/welcome.rs
@@ -32,6 +32,7 @@ const PROVIDER_ENV_VARS: &[(&str, &str)] = &[
     ("ANTHROPIC_API_KEY", "Anthropic"),
     ("OPENAI_API_KEY", "OpenAI"),
     ("DEEPSEEK_API_KEY", "DeepSeek"),
+    ("VOLCENGINE_API_KEY", "Volcano Engine"),
     ("GEMINI_API_KEY", "Gemini"),
     ("GOOGLE_API_KEY", "Gemini"),
     ("GROQ_API_KEY", "Groq"),
@@ -47,7 +48,7 @@ const PROVIDER_ENV_VARS: &[(&str, &str)] = &[
 /// Returns (provider_name, env_var_name) for the first detected key, or None.
 fn detect_provider() -> Option<(&'static str, &'static str)> {
     for &(var, name) in PROVIDER_ENV_VARS {
-        if std::env::var(var).is_ok() {
+        if std::env::var(var).ok().filter(|v| !v.is_empty()).is_some() {
             return Some((name, var));
         }
     }
diff --git a/crates/openfang-cli/src/tui/screens/wizard.rs b/crates/openfang-cli/src/tui/screens/wizard.rs
index f15b8f8c8..27b6ae782 100644
--- a/crates/openfang-cli/src/tui/screens/wizard.rs
+++ b/crates/openfang-cli/src/tui/screens/wizard.rs
@@ -13,6 +13,7 @@ use crate::tui::theme;
 /// Provider metadata for the setup wizard.
 struct ProviderInfo {
     name: &'static str,
+    display: &'static str,
     env_var: &'static str,
     default_model: &'static str,
     needs_key: bool,
@@ -21,108 +22,140 @@ struct ProviderInfo {
 const PROVIDERS: &[ProviderInfo] = &[
     ProviderInfo {
         name: "groq",
+        display: "Groq",
         env_var: "GROQ_API_KEY",
         default_model: "llama-3.3-70b-versatile",
         needs_key: true,
     },
     ProviderInfo {
         name: "anthropic",
+        display: "Anthropic",
         env_var: "ANTHROPIC_API_KEY",
         default_model: "claude-sonnet-4-20250514",
         needs_key: true,
     },
     ProviderInfo {
         name: "openai",
+        display: "OpenAI",
         env_var: "OPENAI_API_KEY",
         default_model: "gpt-4o",
         needs_key: true,
     },
     ProviderInfo {
         name: "openrouter",
+        display: "OpenRouter",
         env_var: "OPENROUTER_API_KEY",
         default_model: "google/gemini-2.5-flash",
         needs_key: true,
     },
     ProviderInfo {
         name: "deepseek",
+        display: "DeepSeek",
         env_var: "DEEPSEEK_API_KEY",
         default_model: "deepseek-chat",
         needs_key: true,
     },
     ProviderInfo {
         name: "together",
+        display: "Together AI",
         env_var: "TOGETHER_API_KEY",
         default_model: "meta-llama/Llama-3.3-70B-Instruct-Turbo",
         needs_key: true,
     },
     ProviderInfo {
         name: "mistral",
+        display: "Mistral",
         env_var: "MISTRAL_API_KEY",
         default_model: "mistral-large-latest",
         needs_key: true,
     },
     ProviderInfo {
         name: "fireworks",
+        display: "Fireworks AI",
         env_var: "FIREWORKS_API_KEY",
         default_model: "accounts/fireworks/models/llama-v3p3-70b-instruct",
         needs_key: true,
     },
     ProviderInfo {
         name: "gemini",
+        display: "Gemini",
         env_var: "GEMINI_API_KEY",
         default_model: "gemini-2.5-flash",
         needs_key: true,
     },
     ProviderInfo {
         name: "xai",
+        display: "xAI",
         env_var: "XAI_API_KEY",
         default_model: "grok-4-0709",
         needs_key: true,
     },
     ProviderInfo {
         name: "qwen",
+        display: "Qwen",
         env_var: "DASHSCOPE_API_KEY",
         default_model: "qwen-plus",
         needs_key: true,
     },
+    ProviderInfo {
+        name: "volcengine",
+        display: "Volcano Engine",
+        env_var: "VOLCENGINE_API_KEY",
+        default_model: "doubao-seed-1-6-251015",
+        needs_key: true,
+    },
+    ProviderInfo {
+        name: "volcengine_coding",
+        display: "Volcano Engine (Coding Plan)",
+        env_var: "VOLCENGINE_API_KEY",
+        default_model: "ark-code-latest",
+        needs_key: true,
+    },
     ProviderInfo {
         name: "perplexity",
+        display: "Perplexity",
         env_var: "PERPLEXITY_API_KEY",
         default_model: "sonar-pro",
         needs_key: true,
     },
     ProviderInfo {
         name: "cohere",
-        env_var: "CO_API_KEY",
+        display: "Cohere",
+        env_var: "COHERE_API_KEY",
         default_model: "command-a",
         needs_key: true,
     },
     ProviderInfo {
         name: "cerebras",
+        display: "Cerebras",
         env_var: "CEREBRAS_API_KEY",
         default_model: "llama-3.3-70b",
         needs_key: true,
     },
     ProviderInfo {
         name: "sambanova",
+        display: "SambaNova",
         env_var: "SAMBANOVA_API_KEY",
         default_model: "Meta-Llama-3.3-70B-Instruct",
         needs_key: true,
     },
     ProviderInfo {
         name: "moonshot",
+        display: "Moonshot",
         env_var: "MOONSHOT_API_KEY",
         default_model: "moonshot-v1-128k",
         needs_key: true,
     },
     ProviderInfo {
         name: "zhipu",
+        display: "Zhipu AI",
         env_var: "ZHIPU_API_KEY",
         default_model: "glm-4-plus",
         needs_key: true,
     },
     ProviderInfo {
         name: "zhipu_coding",
+        display: "Zhipu AI (Coding)",
         env_var: "ZHIPU_API_KEY",
         default_model: "codegeex-4",
         needs_key: true,
@@ -135,24 +168,28 @@ const PROVIDERS: &[ProviderInfo] = &[
     },
     ProviderInfo {
         name: "claude-code",
+        display: "Claude Code",
         env_var: "",
         default_model: "claude-code/sonnet",
         needs_key: false,
     },
     ProviderInfo {
         name: "ollama",
+        display: "Ollama",
         env_var: "OLLAMA_API_KEY",
         default_model: "llama3.2",
         needs_key: false,
     },
     ProviderInfo {
         name: "vllm",
+        display: "vLLM",
         env_var: "VLLM_API_KEY",
         default_model: "local-model",
         needs_key: false,
     },
     ProviderInfo {
         name: "lmstudio",
+        display: "LM Studio",
         env_var: "LMSTUDIO_API_KEY",
         default_model: "local-model",
         needs_key: false,
@@ -544,7 +581,7 @@ fn draw_provider(f: &mut Frame, area: Rect, state: &mut WizardState) {
                 format!("requires {}", p.env_var)
             };
             ListItem::new(Line::from(vec![
-                Span::raw(format!("  {:<14}", p.name)),
+                Span::raw(format!("  {:<24}", p.display)),
                 Span::styled(hint, theme::dim_style()),
             ]))
         })
diff --git a/crates/openfang-kernel/src/kernel.rs b/crates/openfang-kernel/src/kernel.rs
index f449addac..6071a830a 100644
--- a/crates/openfang-kernel/src/kernel.rs
+++ b/crates/openfang-kernel/src/kernel.rs
@@ -2043,12 +2043,22 @@ impl OpenFangKernel {
 
                     // Persist usage to database (same as non-streaming path)
                     let model = &manifest.model.model;
+                    // Reconstruct canonical catalog ID (e.g. "alibaba-coding-plan/qwen3.5-plus")
+                    // when the provider prefix was stripped before the agent ran, so that
+                    // catalog.pricing() finds the correct $0 subscription entry.
+                    let catalog_model_id = {
+                        let cat = kernel_clone
+                            .model_catalog
+                            .read()
+                            .unwrap_or_else(|e| e.into_inner());
+                        Self::resolve_catalog_model_id(&manifest.model.provider, model, &cat)
+                    };
                     let cost = MeteringEngine::estimate_cost_with_catalog(
                         &kernel_clone
                             .model_catalog
                             .read()
                             .unwrap_or_else(|e| e.into_inner()),
-                        model,
+                        &catalog_model_id,
                         result.total_usage.input_tokens,
                         result.total_usage.output_tokens,
                     );
@@ -2600,9 +2610,16 @@ impl OpenFangKernel {
 
         // Record usage in the metering engine (uses catalog pricing as single source of truth)
         let model = &manifest.model.model;
+        // The model field may have had its provider prefix stripped (e.g. "qwen3.5-plus" instead
+        // of "alibaba-coding-plan/qwen3.5-plus"). Reconstruct the canonical catalog ID so that
+        // catalog.pricing() can find the entry and return the correct $0 subscription rate.
+        let catalog_model_id = {
+            let cat = self.model_catalog.read().unwrap_or_else(|e| e.into_inner());
+            Self::resolve_catalog_model_id(&manifest.model.provider, model, &cat)
+        };
         let cost = MeteringEngine::estimate_cost_with_catalog(
             &self.model_catalog.read().unwrap_or_else(|e| e.into_inner()),
-            model,
+            &catalog_model_id,
             result.total_usage.input_tokens,
             result.total_usage.output_tokens,
         );
@@ -2634,6 +2651,31 @@ impl OpenFangKernel {
         Ok(result)
     }
 
+    /// Reconstructs the canonical catalog model ID for any provider that stores models
+    /// with a `provider/model` prefix format (alibaba-coding-plan, codex, copilot,
+    /// qwen-code, etc.). Used before metering to ensure catalog lookup succeeds.
+    ///
+    /// When a provider prefix was stripped before the agent ran (e.g. "qwen3.5-plus"
+    /// instead of "alibaba-coding-plan/qwen3.5-plus"), this helper rebuilds the
+    /// prefixed form so that `catalog.pricing()` can find the correct entry.
+    /// If the model already contains '/' it is returned as-is.
+    fn resolve_catalog_model_id(
+        provider: &str,
+        model: &str,
+        cat: &openfang_runtime::model_catalog::ModelCatalog,
+    ) -> String {
+        if model.contains('/') {
+            return model.to_owned();
+        }
+        let provider_hyphen = provider.replace('_', "-");
+        let prefixed = format!("{}/{}", provider_hyphen, model);
+        if cat.pricing(&prefixed).is_some() {
+            prefixed
+        } else {
+            model.to_owned()
+        }
+    }
+
     /// Resolve a module path relative to the kernel's home directory.
     ///
     /// If the path is absolute, return it as-is. Otherwise, resolve relative
@@ -3094,9 +3136,13 @@ impl OpenFangKernel {
             .unwrap_or((0, 0));
 
         let model = &entry.manifest.model.model;
+        let catalog_model_id = {
+            let cat = self.model_catalog.read().unwrap_or_else(|e| e.into_inner());
+            Self::resolve_catalog_model_id(&entry.manifest.model.provider, model, &cat)
+        };
         let cost = MeteringEngine::estimate_cost_with_catalog(
             &self.model_catalog.read().unwrap_or_else(|e| e.into_inner()),
-            model,
+            &catalog_model_id,
             input_tokens,
             output_tokens,
         );
@@ -6844,4 +6890,47 @@ mod tests {
 
         kernel.shutdown();
     }
+
+    /// `resolve_catalog_model_id` must work for any provider that stores models with a
+    /// `provider/model` prefix, not just alibaba-coding-plan.
+    #[test]
+    fn test_resolve_catalog_model_id_multi_provider() {
+        use openfang_runtime::model_catalog::ModelCatalog;
+
+        let cat = ModelCatalog::new();
+
+        // Case 1: alibaba-coding-plan — model exists in catalog after prefix rebuild.
+        let result = OpenFangKernel::resolve_catalog_model_id(
+            "alibaba_coding_plan",
+            "qwen3.5-plus",
+            &cat,
+        );
+        assert_eq!(
+            result, "alibaba-coding-plan/qwen3.5-plus",
+            "should rebuild prefixed form when catalog entry exists"
+        );
+
+        // Case 2: model already contains '/' — returned unchanged regardless of provider.
+        let already_prefixed = OpenFangKernel::resolve_catalog_model_id(
+            "alibaba_coding_plan",
+            "alibaba-coding-plan/qwen3.5-plus",
+            &cat,
+        );
+        assert_eq!(
+            already_prefixed, "alibaba-coding-plan/qwen3.5-plus",
+            "already-prefixed model should pass through unchanged"
+        );
+
+        // Case 3: unknown provider (e.g. qwen_code) with no catalog entry — falls back
+        // to the bare model name so callers can still proceed.
+        let fallback = OpenFangKernel::resolve_catalog_model_id(
+            "qwen_code",
+            "some-unknown-model",
+            &cat,
+        );
+        assert_eq!(
+            fallback, "some-unknown-model",
+            "when prefixed form is not in catalog, bare model name should be returned"
+        );
+    }
 }
diff --git a/crates/openfang-kernel/src/metering.rs b/crates/openfang-kernel/src/metering.rs
index e34dff4e9..2ad0b6d13 100644
--- a/crates/openfang-kernel/src/metering.rs
+++ b/crates/openfang-kernel/src/metering.rs
@@ -374,6 +374,14 @@ fn estimate_cost_rates(model: &str) -> (f64, f64) {
     if model.contains("llama") || model.contains("mixtral") {
         return (0.05, 0.10);
     }
+    // ── Alibaba Coding Plan (subscription-based: $50/month) ─────
+    // Must come BEFORE the qwen/glm/kimi/minimax checks below
+    // because model IDs like "alibaba-coding-plan/qwen3.5-plus" contain "qwen".
+    // All models in the Coding Plan use flat-rate subscription pricing.
+    // Per-token costs are $0 — actual cost is the fixed monthly fee.
+    if model.contains("alibaba-coding-plan") {
+        return (0.0, 0.0);
+    }
     // ── Qwen (Alibaba) ──────────────────────────────────────────
     if model.contains("qwen-max") {
         return (4.00, 12.00);
@@ -387,6 +395,9 @@ fn estimate_cost_rates(model: &str) -> (f64, f64) {
     if model.contains("qwen-turbo") {
         return (0.30, 0.60);
     }
+    // ── Qwen (generic / pay-per-token) ──────────────────────────
+    // Note: "alibaba-coding-plan/*" models are caught by the earlier guard above
+    // and never reach this branch.
     if model.contains("qwen") {
         return (0.20, 0.60);
     }
@@ -751,6 +762,70 @@ mod tests {
         assert!((cost - 0.12).abs() < 0.01); // $0.06 + $0.06
     }
 
+    #[test]
+    fn test_estimate_cost_alibaba_coding_plan_subscription() {
+        // Alibaba Coding Plan is subscription-based ($50/month) — per-token cost is $0
+        let qwen35_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/qwen3.5-plus",
+            1_000_000,
+            1_000_000,
+        );
+        assert!((qwen35_cost).abs() < 0.001, "Qwen 3.5 Plus should be $0 (subscription)");
+
+        let glm5_cost =
+            MeteringEngine::estimate_cost("alibaba-coding-plan/glm-5", 1_000_000, 1_000_000);
+        assert!((glm5_cost).abs() < 0.001, "GLM-5 should be $0 (subscription)");
+
+        let kimi_cost =
+            MeteringEngine::estimate_cost("alibaba-coding-plan/kimi-k2.5", 1_000_000, 1_000_000);
+        assert!((kimi_cost).abs() < 0.001, "Kimi K2.5 should be $0 (subscription)");
+
+        let minimax_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/minimax-m2.5",
+            1_000_000,
+            1_000_000,
+        );
+        assert!((minimax_cost).abs() < 0.001, "MiniMax M2.5 should be $0 (subscription)");
+
+        let glm47_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/glm-4.7",
+            1_000_000,
+            1_000_000,
+        );
+        assert!((glm47_cost).abs() < 0.001, "GLM-4.7 should be $0 (subscription)");
+
+        let qwen3_max_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/qwen3-max-2026-01-23",
+            1_000_000,
+            1_000_000,
+        );
+        assert!(
+            (qwen3_max_cost).abs() < 0.001,
+            "Qwen3-max-2026-01-23 should be $0 (subscription)"
+        );
+
+        let qwen3_coder_plus_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/qwen3-coder-plus",
+            1_000_000,
+            1_000_000,
+        );
+        assert!(
+            (qwen3_coder_plus_cost).abs() < 0.001,
+            "Qwen3-coder-plus should be $0 (subscription)"
+        );
+
+        let qwen3_coder_next_cost = MeteringEngine::estimate_cost(
+            "alibaba-coding-plan/qwen3-coder-next",
+            1_000_000,
+            1_000_000,
+        );
+        assert!(
+            (qwen3_coder_next_cost).abs() < 0.001,
+            "Qwen3-coder-next should be $0 (subscription)"
+        );
+
+    }
+
     #[test]
     fn test_estimate_cost_with_catalog() {
         let catalog = openfang_runtime::model_catalog::ModelCatalog::new();
diff --git a/crates/openfang-runtime/src/agent_loop.rs b/crates/openfang-runtime/src/agent_loop.rs
index f773def41..065d2af77 100644
--- a/crates/openfang-runtime/src/agent_loop.rs
+++ b/crates/openfang-runtime/src/agent_loop.rs
@@ -18,6 +18,9 @@ use openfang_memory::session::Session;
 use openfang_memory::MemorySubstrate;
 use openfang_skills::registry::SkillRegistry;
 use openfang_types::agent::{AgentManifest, FallbackModel};
+use openfang_types::model_catalog::{
+    DOUBAO_PROVIDER_ID, VOLCENGINE_CODING_PROVIDER_ID, VOLCENGINE_PROVIDER_ID,
+};
 use openfang_types::error::{OpenFangError, OpenFangResult};
 use openfang_types::memory::{Memory, MemoryFilter, MemorySource};
 use openfang_types::message::{
@@ -107,15 +110,30 @@ fn append_tool_error_guidance(tool_result_blocks: &mut Vec<ContentBlock>) {
 /// Many models are stored as `provider/org/model` (e.g. `openrouter/google/gemini-2.5-flash`)
 /// but the upstream API expects just `org/model` (e.g. `google/gemini-2.5-flash`).
 pub fn strip_provider_prefix(model: &str, provider: &str) -> String {
-    let slash_prefix = format!("{}/", provider);
-    let colon_prefix = format!("{}:", provider);
-    if model.starts_with(&slash_prefix) {
+    let provider_normalized = provider.replace('_', "-");
+    let slash_prefix = format!("{}/", provider_normalized);
+    let colon_prefix = format!("{}:", provider_normalized);
+    let mut result = if model.starts_with(&slash_prefix) {
         model[slash_prefix.len()..].to_string()
     } else if model.starts_with(&colon_prefix) {
         model[colon_prefix.len()..].to_string()
     } else {
         model.to_string()
+    };
+    // Strip "ark/" catalog namespace prefix before sending to Ark API.
+    // "ark/" is used internally to disambiguate Ark marketplace models from
+    // native provider models with the same name (e.g. ark/minimax-m2.5 vs
+    // minimax provider's minimax-m2.5). The Ark API endpoint expects the bare
+    // model name (e.g. "minimax-m2.5"), not the namespaced form.
+    // Strip ark/ prefix only for Volcano Engine providers (Ark marketplace models)
+    if (provider == VOLCENGINE_CODING_PROVIDER_ID
+        || provider == VOLCENGINE_PROVIDER_ID
+        || provider == DOUBAO_PROVIDER_ID)
+        && result.starts_with("ark/")
+    {
+        result = result["ark/".len()..].to_string();
     }
+    result
 }
 
 /// Default context window size (tokens) for token-based trimming.
@@ -2953,6 +2971,30 @@ mod tests {
     use openfang_types::tool::ToolCall;
     use std::sync::atomic::{AtomicU32, Ordering};
 
+    #[test]
+    fn test_strip_provider_prefix_alibaba() {
+        // alibaba: underscore provider normalizes to hyphen prefix
+        assert_eq!(
+            strip_provider_prefix("alibaba-coding-plan/qwen3.5-plus", "alibaba_coding_plan"),
+            "qwen3.5-plus"
+        );
+        // no prefix passthrough
+        assert_eq!(
+            strip_provider_prefix("qwen3.5-plus", "alibaba_coding_plan"),
+            "qwen3.5-plus"
+        );
+        // regression: kimi_coding model passes through unchanged
+        assert_eq!(
+            strip_provider_prefix("kimi-for-coding", "kimi_coding"),
+            "kimi-for-coding"
+        );
+        // regression: zhipu_coding model passes through unchanged
+        assert_eq!(
+            strip_provider_prefix("codegeex-4", "zhipu_coding"),
+            "codegeex-4"
+        );
+    }
+
     #[test]
     fn test_max_iterations_constant() {
         assert_eq!(MAX_ITERATIONS, 50);
@@ -3022,6 +3064,42 @@ mod tests {
         assert_eq!(MAX_HISTORY_MESSAGES, 20);
     }
 
+    #[test]
+    fn test_strip_ark_catalog_prefix_for_volcengine_coding() {
+        // ark/ is catalog-only; Ark API expects bare name
+        assert_eq!(
+            strip_provider_prefix("ark/doubao-seed-code", "volcengine_coding"),
+            "doubao-seed-code"
+        );
+    }
+
+    #[test]
+    fn test_strip_provider_prefix_ark_volcengine() {
+        // Should strip ark/ for volcengine
+        assert_eq!(
+            strip_provider_prefix("ark/doubao-seed-code", "volcengine"),
+            "doubao-seed-code"
+        );
+    }
+
+    #[test]
+    fn test_strip_provider_prefix_ark_doubao() {
+        // Should strip ark/ for doubao provider alias
+        assert_eq!(
+            strip_provider_prefix("ark/some-model", "doubao"),
+            "some-model"
+        );
+    }
+
+    #[test]
+    fn test_strip_provider_prefix_ark_not_stripped_for_other_providers() {
+        // Must NOT strip ark/ for non-volcengine providers
+        assert_eq!(
+            strip_provider_prefix("ark/some-model", "openai"),
+            "ark/some-model"
+        );
+    }
+
     // --- Integration tests for empty response guards ---
 
     fn test_manifest() -> AgentManifest {
diff --git a/crates/openfang-runtime/src/drivers/mod.rs b/crates/openfang-runtime/src/drivers/mod.rs
index 2df8923d4..13a6185b2 100644
--- a/crates/openfang-runtime/src/drivers/mod.rs
+++ b/crates/openfang-runtime/src/drivers/mod.rs
@@ -2,7 +2,8 @@
 //!
 //! Contains drivers for Anthropic Claude, Google Gemini, OpenAI-compatible APIs, and more.
 //! Supports: Anthropic, Gemini, OpenAI, Groq, OpenRouter, DeepSeek, Together,
-//! Mistral, Fireworks, Ollama, vLLM, Chutes.ai, and any OpenAI-compatible endpoint.
+//! Mistral, Fireworks, Ollama, vLLM, Chutes.ai, volcengine (Doubao / Ark),
+//! and any OpenAI-compatible endpoint.
 
 pub mod anthropic;
 pub mod claude_code;
@@ -15,14 +16,15 @@ pub mod vertex;
 
 use crate::llm_driver::{DriverConfig, LlmDriver, LlmError};
 use openfang_types::model_catalog::{
-    AI21_BASE_URL, ANTHROPIC_BASE_URL, AZURE_OPENAI_BASE_URL, CEREBRAS_BASE_URL, CHUTES_BASE_URL,
-    COHERE_BASE_URL, DEEPSEEK_BASE_URL, FIREWORKS_BASE_URL, GEMINI_BASE_URL, GROQ_BASE_URL,
-    HUGGINGFACE_BASE_URL, KIMI_CODING_BASE_URL, LEMONADE_BASE_URL, LMSTUDIO_BASE_URL,
-    MINIMAX_BASE_URL, MISTRAL_BASE_URL, MOONSHOT_BASE_URL, NVIDIA_NIM_BASE_URL, OLLAMA_BASE_URL,
-    OPENAI_BASE_URL, OPENROUTER_BASE_URL, PERPLEXITY_BASE_URL, QIANFAN_BASE_URL, QWEN_BASE_URL,
-    REPLICATE_BASE_URL, SAMBANOVA_BASE_URL, TOGETHER_BASE_URL, VENICE_BASE_URL, VLLM_BASE_URL,
-    VOLCENGINE_BASE_URL, VOLCENGINE_CODING_BASE_URL, XAI_BASE_URL, ZAI_BASE_URL,
-    ZAI_CODING_BASE_URL, ZHIPU_BASE_URL, ZHIPU_CODING_BASE_URL,
+    AI21_BASE_URL, ALIBABA_CODING_PLAN_BASE_URL, ANTHROPIC_BASE_URL, AZURE_OPENAI_BASE_URL,
+    CEREBRAS_BASE_URL, CHUTES_BASE_URL, COHERE_BASE_URL, DEEPSEEK_BASE_URL, FIREWORKS_BASE_URL,
+    GEMINI_BASE_URL, GROQ_BASE_URL, HUGGINGFACE_BASE_URL, KIMI_CODING_BASE_URL,
+    LEMONADE_BASE_URL, LMSTUDIO_BASE_URL, MINIMAX_BASE_URL, MISTRAL_BASE_URL, MOONSHOT_BASE_URL,
+    NVIDIA_NIM_BASE_URL, OLLAMA_BASE_URL, OPENAI_BASE_URL, OPENROUTER_BASE_URL,
+    PERPLEXITY_BASE_URL, QIANFAN_BASE_URL, QWEN_BASE_URL, REPLICATE_BASE_URL, SAMBANOVA_BASE_URL,
+    TOGETHER_BASE_URL, VENICE_BASE_URL, VLLM_BASE_URL, VOLCENGINE_BASE_URL,
+    VOLCENGINE_CODING_BASE_URL, XAI_BASE_URL, ZAI_BASE_URL, ZAI_CODING_BASE_URL,
+    ZHIPU_BASE_URL, ZHIPU_CODING_BASE_URL,
 };
 use std::sync::Arc;
 
@@ -197,6 +199,7 @@ fn provider_defaults(provider: &str) -> Option<ProviderDefaults> {
             api_key_env: "QIANFAN_API_KEY",
             key_required: true,
         }),
+        // "doubao" is also a model alias in builtin_aliases; here it acts as a provider alias
         "volcengine" | "doubao" => Some(ProviderDefaults {
             base_url: VOLCENGINE_BASE_URL,
             api_key_env: "VOLCENGINE_API_KEY",
@@ -260,6 +263,7 @@ fn provider_defaults(provider: &str) -> Option<ProviderDefaults> {
 /// - `xai` — xAI (Grok)
 /// - `replicate` — Replicate
 /// - `chutes` — Chutes.ai (serverless open-source model inference)
+/// - `volcengine` — Volcano Engine (Doubao/Ark)
 /// - Any custom provider with `base_url` set uses OpenAI-compatible format
 pub fn create_driver(config: &DriverConfig) -> Result<Arc<dyn LlmDriver>, LlmError> {
     let provider = config.provider.as_str();
@@ -426,6 +430,29 @@ pub fn create_driver(config: &DriverConfig) -> Result<Arc<dyn LlmDriver>, LlmErr
         return Ok(Arc::new(anthropic::AnthropicDriver::new(api_key, base_url)));
     }
 
+    // Alibaba Coding Plan — OpenAI-compatible endpoint.
+    // Accept both underscore form ("alibaba_coding_plan") and hyphen form ("alibaba-coding-plan").
+    // NOTE: prefix stripping ("alibaba-coding-plan/") is handled upstream by
+    // `strip_provider_prefix` in agent_loop.rs before the CompletionRequest is built.
+    // By the time the request reaches this driver, `request.model` is already the bare
+    // model name (e.g. "qwen3.5-plus"). No wrapper needed here.
+    if provider == "alibaba_coding_plan" || provider == "alibaba-coding-plan" {
+        let api_key = config
+            .api_key
+            .clone()
+            .or_else(|| std::env::var("ALIBABA_CODING_PLAN_API_KEY").ok())
+            .ok_or_else(|| {
+                LlmError::MissingApiKey(
+                    "Set ALIBABA_CODING_PLAN_API_KEY environment variable".to_string(),
+                )
+            })?;
+        let base_url = config
+            .base_url
+            .clone()
+            .unwrap_or_else(|| ALIBABA_CODING_PLAN_BASE_URL.to_string());
+        return Ok(Arc::new(openai::OpenAIDriver::new(api_key, base_url)));
+    }
+
     // All other providers use OpenAI-compatible format
     if let Some(defaults) = provider_defaults(provider) {
         let api_key = config
@@ -489,7 +516,11 @@ pub fn create_driver(config: &DriverConfig) -> Result<Arc<dyn LlmDriver>, LlmErr
             "Unknown provider '{}'. Supported: anthropic, gemini, openai, azure, groq, openrouter, \
              deepseek, together, mistral, fireworks, ollama, vllm, lmstudio, perplexity, \
              cohere, ai21, cerebras, sambanova, huggingface, xai, replicate, github-copilot, \
-             chutes, venice, nvidia, codex, claude-code. Or set base_url for a custom OpenAI-compatible endpoint.",
+             chutes, venice, nvidia, codex, claude-code, qwen-code, kimi_coding, zhipu_coding, \
+             zai, zai_coding, qianfan, moonshot, lemonade, \
+             alibaba_coding_plan (or alibaba-coding-plan), \
+             volcengine (or doubao), volcengine_coding. \
+             Or set base_url for a custom OpenAI-compatible endpoint.",
             provider
         ),
     })
@@ -507,6 +538,7 @@ pub fn detect_available_provider() -> Option<(&'static str, &'static str, &'stat
         ("gemini", "gemini-2.5-flash", "GEMINI_API_KEY"),
         ("groq", "llama-3.3-70b-versatile", "GROQ_API_KEY"),
         ("deepseek", "deepseek-chat", "DEEPSEEK_API_KEY"),
+        ("volcengine", "doubao-seed-1-6-251015", "VOLCENGINE_API_KEY"),
         (
             "openrouter",
             "openrouter/google/gemini-2.5-flash",
@@ -530,6 +562,11 @@ pub fn detect_available_provider() -> Option<(&'static str, &'static str, &'stat
             "PERPLEXITY_API_KEY",
         ),
         ("cohere", "command-r-plus", "COHERE_API_KEY"),
+        (
+            "alibaba_coding_plan",
+            "alibaba-coding-plan/qwen3.5-plus",
+            "ALIBABA_CODING_PLAN_API_KEY",
+        ),
     ];
     for &(provider, model, env_var) in PROBE_ORDER {
         if std::env::var(env_var)
@@ -584,6 +621,8 @@ pub fn known_providers() -> &'static [&'static str] {
         "kimi_coding",
         "qianfan",
         "volcengine",
+        "volcengine_coding",
+        "doubao",
         "chutes",
         "venice",
         "nvidia",
@@ -591,6 +630,7 @@ pub fn known_providers() -> &'static [&'static str] {
         "claude-code",
         "qwen-code",
         "azure",
+        "alibaba_coding_plan",
     ]
 }
 
@@ -689,13 +729,17 @@ mod tests {
         assert!(providers.contains(&"kimi_coding"));
         assert!(providers.contains(&"qianfan"));
         assert!(providers.contains(&"volcengine"));
+        assert!(providers.contains(&"volcengine_coding"));
+        assert!(providers.contains(&"doubao"));
         assert!(providers.contains(&"chutes"));
         assert!(providers.contains(&"nvidia"));
         assert!(providers.contains(&"codex"));
         assert!(providers.contains(&"claude-code"));
         assert!(providers.contains(&"qwen-code"));
         assert!(providers.contains(&"azure"));
-        assert_eq!(providers.len(), 37);
+        // Alibaba Coding Plan
+        assert!(providers.contains(&"alibaba_coding_plan"));
+        assert_eq!(providers.len(), 40);
     }
 
     #[test]
@@ -768,6 +812,93 @@ mod tests {
         assert!(driver.is_err());
     }
 
+    #[test]
+    fn test_alibaba_coding_plan_driver_with_direct_key() {
+        // Test that a directly-supplied API key is accepted.
+        let config = DriverConfig {
+            provider: "alibaba_coding_plan".to_string(),
+            api_key: Some("sk-sp-test-direct-key-67890".to_string()),
+            base_url: None,
+            skip_permissions: true,
+        };
+        let driver = create_driver(&config);
+        assert!(
+            driver.is_ok(),
+            "Alibaba Coding Plan with direct API key should succeed"
+        );
+    }
+
+    #[test]
+    fn test_alibaba_coding_plan_no_key_errors() {
+        // Alibaba Coding Plan with no API key should return MissingApiKey.
+        // We supply api_key: None and a unique base_url so the alibaba_coding_plan
+        // branch is entered deterministically without reading or modifying any env var.
+        // This is safe to run in parallel with other tests.
+        let config = DriverConfig {
+            provider: "alibaba_coding_plan".to_string(),
+            api_key: None,
+            base_url: Some("https://test-alibaba-no-key.invalid".to_string()),
+            skip_permissions: true,
+        };
+        let driver = create_driver(&config);
+        assert!(driver.is_err(), "Expected error when api_key is None");
+        let err = driver.err().unwrap().to_string();
+        assert!(
+            err.contains("ALIBABA_CODING_PLAN_API_KEY"),
+            "Error should mention ALIBABA_CODING_PLAN_API_KEY: {}",
+            err
+        );
+    }
+
+    #[test]
+    fn test_alibaba_coding_plan_custom_base_url() {
+        // Test custom base URL override
+        let config = DriverConfig {
+            provider: "alibaba_coding_plan".to_string(),
+            api_key: Some("sk-sp-test-key".to_string()),
+            base_url: Some("https://custom-endpoint.example.com/v1".to_string()),
+            skip_permissions: true,
+        };
+        let driver = create_driver(&config);
+        assert!(driver.is_ok());
+    }
+
+    #[test]
+    fn test_doubao_alias_driver_with_direct_key() {
+        // "doubao" is a provider alias for volcengine — create_driver should route it correctly.
+        let config = DriverConfig {
+            provider: "doubao".to_string(),
+            api_key: Some("sk-test-doubao-direct-key-12345".to_string()),
+            base_url: None,
+            skip_permissions: true,
+        };
+        let driver = create_driver(&config);
+        assert!(
+            driver.is_ok(),
+            "doubao provider alias with direct API key should succeed: {:?}",
+            driver.err()
+        );
+    }
+
+    #[test]
+    fn test_doubao_alias_driver_no_key_errors() {
+        // "doubao" with no key and a unique base_url should return MissingApiKey.
+        let config = DriverConfig {
+            provider: "doubao".to_string(),
+            api_key: None,
+            base_url: Some("https://test-doubao-no-key.invalid".to_string()),
+            skip_permissions: true,
+        };
+        let driver = create_driver(&config);
+        assert!(driver.is_err(), "Expected error when doubao api_key is None");
+        let err = driver.err().unwrap().to_string();
+        assert!(
+            err.contains("VOLCENGINE_API_KEY"),
+            "Error should mention VOLCENGINE_API_KEY: {}",
+            err
+        );
+    }
+
     #[test]
     fn test_custom_provider_key_no_url_helpful_error() {
         // Custom provider with key set (via env) but no base_url should give helpful error.
@@ -888,4 +1019,31 @@ mod tests {
             "azure-openai alias should create driver successfully"
         );
     }
+
+    #[test]
+    fn test_provider_defaults_volcengine() {
+        let d = provider_defaults("volcengine").unwrap();
+        assert_eq!(d.base_url, "https://ark.cn-beijing.volces.com/api/v3");
+        assert_eq!(d.api_key_env, "VOLCENGINE_API_KEY");
+        assert!(d.key_required);
+    }
+
+    #[test]
+    fn test_provider_defaults_volcengine_doubao_alias() {
+        let d = provider_defaults("doubao").unwrap();
+        assert_eq!(d.base_url, "https://ark.cn-beijing.volces.com/api/v3");
+        assert_eq!(d.api_key_env, "VOLCENGINE_API_KEY");
+        assert!(d.key_required);
+    }
+
+    #[test]
+    fn test_provider_defaults_volcengine_coding() {
+        let d = provider_defaults("volcengine_coding").unwrap();
+        assert_eq!(
+            d.base_url,
+            "https://ark.cn-beijing.volces.com/api/coding/v3"
+        );
+        assert_eq!(d.api_key_env, "VOLCENGINE_API_KEY");
+        assert!(d.key_required);
+    }
 }
diff --git a/crates/openfang-runtime/src/model_catalog.rs b/crates/openfang-runtime/src/model_catalog.rs
index 62b81c74e..fff666b9f 100644
--- a/crates/openfang-runtime/src/model_catalog.rs
+++ b/crates/openfang-runtime/src/model_catalog.rs
@@ -4,15 +4,16 @@
 //! with alias resolution, auth status detection, and pricing lookups.
 
 use openfang_types::model_catalog::{
-    AuthStatus, ModelCatalogEntry, ModelTier, ProviderInfo, AI21_BASE_URL, ANTHROPIC_BASE_URL,
-    AZURE_OPENAI_BASE_URL, BEDROCK_BASE_URL, CEREBRAS_BASE_URL, CHUTES_BASE_URL, COHERE_BASE_URL,
-    DEEPSEEK_BASE_URL, FIREWORKS_BASE_URL, GEMINI_BASE_URL, GITHUB_COPILOT_BASE_URL, GROQ_BASE_URL,
-    HUGGINGFACE_BASE_URL, KIMI_CODING_BASE_URL, LEMONADE_BASE_URL, LMSTUDIO_BASE_URL,
-    MINIMAX_BASE_URL, MISTRAL_BASE_URL, MOONSHOT_BASE_URL, NVIDIA_NIM_BASE_URL, OLLAMA_BASE_URL,
-    OPENAI_BASE_URL, OPENROUTER_BASE_URL, PERPLEXITY_BASE_URL, QIANFAN_BASE_URL, QWEN_BASE_URL,
-    REPLICATE_BASE_URL, SAMBANOVA_BASE_URL, TOGETHER_BASE_URL, VENICE_BASE_URL, VLLM_BASE_URL,
-    VOLCENGINE_BASE_URL, VOLCENGINE_CODING_BASE_URL, XAI_BASE_URL, ZAI_BASE_URL,
-    ZAI_CODING_BASE_URL, ZHIPU_BASE_URL, ZHIPU_CODING_BASE_URL,
+    AuthStatus, ModelCatalogEntry, ModelTier, ProviderInfo, AI21_BASE_URL,
+    ALIBABA_CODING_PLAN_BASE_URL, ANTHROPIC_BASE_URL, AZURE_OPENAI_BASE_URL, BEDROCK_BASE_URL,
+    CEREBRAS_BASE_URL, CHUTES_BASE_URL, COHERE_BASE_URL, DEEPSEEK_BASE_URL, FIREWORKS_BASE_URL,
+    GEMINI_BASE_URL, GITHUB_COPILOT_BASE_URL, GROQ_BASE_URL, HUGGINGFACE_BASE_URL,
+    KIMI_CODING_BASE_URL, LEMONADE_BASE_URL, LMSTUDIO_BASE_URL, MINIMAX_BASE_URL, MISTRAL_BASE_URL,
+    MOONSHOT_BASE_URL, NVIDIA_NIM_BASE_URL, OLLAMA_BASE_URL, OPENAI_BASE_URL, OPENROUTER_BASE_URL,
+    PERPLEXITY_BASE_URL, QIANFAN_BASE_URL, QWEN_BASE_URL, REPLICATE_BASE_URL, SAMBANOVA_BASE_URL,
+    TOGETHER_BASE_URL, VENICE_BASE_URL, VLLM_BASE_URL, VOLCENGINE_BASE_URL,
+    VOLCENGINE_CODING_BASE_URL, XAI_BASE_URL, ZAI_BASE_URL, ZAI_CODING_BASE_URL, ZHIPU_BASE_URL,
+    ZHIPU_CODING_BASE_URL,
 };
 use std::collections::HashMap;
 
@@ -81,13 +82,23 @@ impl ModelCatalog {
             }
 
             // Primary: check the provider's declared env var
-            let has_key = std::env::var(&provider.api_key_env).is_ok();
+            let has_key = std::env::var(&provider.api_key_env)
+                .ok()
+                .filter(|v| !v.is_empty())
+                .is_some();
 
             // Secondary: provider-specific fallback auth
             let has_fallback = match provider.id.as_str() {
-                "gemini" => std::env::var("GOOGLE_API_KEY").is_ok(),
+                "gemini" => std::env::var("GOOGLE_API_KEY")
+                    .ok()
+                    .filter(|v| !v.is_empty())
+                    .is_some(),
                 "codex" => {
-                    std::env::var("OPENAI_API_KEY").is_ok() || read_codex_credential().is_some()
+                    std::env::var("OPENAI_API_KEY")
+                        .ok()
+                        .filter(|v| !v.is_empty())
+                        .is_some()
+                        || read_codex_credential().is_some()
                 }
                 // claude-code is handled above (before key_required check)
                 _ => false,
@@ -765,7 +776,7 @@ fn builtin_providers() -> Vec<ProviderInfo> {
             auth_status: AuthStatus::Missing,
             model_count: 0,
         },
-        // ── Chinese providers (5) ────────────────────────────────────
+        // ── Chinese providers (6) ────────────────────────────────────
         ProviderInfo {
             id: "qwen".into(),
             display_name: "Qwen (Alibaba)".into(),
@@ -775,6 +786,17 @@ fn builtin_providers() -> Vec<ProviderInfo> {
             auth_status: AuthStatus::Missing,
             model_count: 0,
         },
+        // Provider ID uses underscores ("alibaba_coding_plan") while model IDs use hyphens ("alibaba-coding-plan/...").
+        // This matches the pattern of zhipu_coding, kimi_coding, etc. Do not normalize — create_driver matches on underscore form.
+        ProviderInfo {
+            id: "alibaba_coding_plan".into(),
+            display_name: "Alibaba Coding Plan (Intl)".into(),
+            api_key_env: "ALIBABA_CODING_PLAN_API_KEY".into(),
+            base_url: ALIBABA_CODING_PLAN_BASE_URL.into(),
+            key_required: true,
+            auth_status: AuthStatus::Missing,
+            model_count: 0,
+        },
         ProviderInfo {
             id: "minimax".into(),
             display_name: "MiniMax".into(),
@@ -850,7 +872,7 @@ fn builtin_providers() -> Vec<ProviderInfo> {
         // ── Volcano Engine (Doubao) ──────────────────────────────────
         ProviderInfo {
             id: "volcengine".into(),
-            display_name: "Volcano Engine (Doubao)".into(),
+            display_name: "Volcano Engine".into(),
             api_key_env: "VOLCENGINE_API_KEY".into(),
             base_url: VOLCENGINE_BASE_URL.into(),
             key_required: true,
@@ -979,6 +1001,11 @@ fn builtin_aliases() -> HashMap<String, String> {
         ("minimax-highspeed", "MiniMax-M2.5-highspeed"),
         ("minimax-m2.1", "MiniMax-M2.1"),
         ("codegeex", "codegeex-4"),
+        // Alibaba Coding Plan aliases
+        ("alibaba-coding-plan", "alibaba-coding-plan/qwen3.5-plus"),
+        // "qwen3-coder" (with digit 3) → alibaba-coding-plan coding endpoint
+        // "qwen-coder"  (no digit)     → qwen-code/qwen3-coder (direct Qwen Code model)
+        ("qwen3-coder", "alibaba-coding-plan/qwen3-coder-plus"),
         // Codex aliases
         ("codex", "codex/gpt-5.4"),
         ("codex-5.4", "codex/gpt-5.4"),
@@ -3193,6 +3220,138 @@ fn builtin_models() -> Vec<ModelCatalogEntry> {
             aliases: vec!["abab7".into()],
         },
         // ══════════════════════════════════════════════════════════════
+        // Alibaba Coding Plan International (8)
+        // All pricing set to $0 — actual cost is fixed monthly subscription.
+        // Provides multi-provider access: Qwen, Zhipu GLM, Moonshot Kimi, MiniMax.
+        // See: https://www.alibabacloud.com/help/en/model-studio/coding-plan
+        // ══════════════════════════════════════════════════════════════
+        // ── Qwen models (4) ──────────────────────────────────────────
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/qwen3.5-plus".into(),
+            display_name: "Qwen 3.5 Plus (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Smart,
+            context_window: 1_000_000,
+            max_output_tokens: 65_536,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            // The Coding Plan endpoint (coding-intl.dashscope.aliyuncs.com) is a
+            // coding-specialized endpoint that does not expose multimodal routes.
+            // Use the standard Qwen provider if vision is required.
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/qwen3-max-2026-01-23".into(),
+            display_name: "Qwen 3 Max 2026-01-23 (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Frontier,
+            context_window: 262_144,
+            max_output_tokens: 65_536,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/qwen3-coder-plus".into(),
+            display_name: "Qwen 3 Coder Plus (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Smart,
+            context_window: 1_000_000,
+            max_output_tokens: 65_536,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/qwen3-coder-next".into(),
+            display_name: "Qwen 3 Coder Next (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Frontier,
+            context_window: 262_144,
+            max_output_tokens: 65_536,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        // ── Zhipu / GLM via Coding Plan (2) ─────────────────────────
+        // API receives "glm-5" / "glm-4.7" after prefix stripping
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/glm-5".into(),
+            display_name: "GLM-5 (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Frontier,
+            context_window: 202_752,
+            max_output_tokens: 32_768,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/glm-4.7".into(),
+            display_name: "GLM-4.7 (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Smart,
+            context_window: 202_752,
+            max_output_tokens: 32_768,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        // ── Moonshot / Kimi via Coding Plan (1) ─────────────────────
+        // API receives "kimi-k2.5" after prefix stripping
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/kimi-k2.5".into(),
+            display_name: "Kimi K2.5 (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Smart,
+            context_window: 262_144,
+            max_output_tokens: 32_768,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            // The Coding Plan endpoint (coding-intl.dashscope.aliyuncs.com) is a
+            // coding-specialized endpoint that does not expose multimodal routes.
+            // Use the standard Kimi/Moonshot provider if vision is required.
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        // ── MiniMax via Coding Plan (1) ──────────────────────────────
+        // The bare alias "minimax-m2.5" is claimed by the MiniMax provider. Users must specify
+        // the full "alibaba-coding-plan/minimax-m2.5" ID to use this model via the Coding Plan endpoint.
+        ModelCatalogEntry {
+            id: "alibaba-coding-plan/minimax-m2.5".into(),
+            display_name: "MiniMax M2.5 (Alibaba Coding Plan)".into(),
+            provider: "alibaba_coding_plan".into(),
+            tier: ModelTier::Smart,
+            context_window: 204_800,
+            max_output_tokens: 32_768,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        // ══════════════════════════════════════════════════════════════
         // Zhipu AI / GLM (6)
         // ══════════════════════════════════════════════════════════════
         ModelCatalogEntry {
@@ -3462,9 +3621,156 @@ fn builtin_models() -> Vec<ModelCatalogEntry> {
             supports_streaming: true,
             aliases: vec![],
         },
+        // ══════════════════════════════════════════════════════════════
+        // Volcano Engine Coding Plan (9)
+        // ══════════════════════════════════════════════════════════════
+        ModelCatalogEntry {
+            id: "ark-code-latest".into(),
+            display_name: "Ark Code (Latest)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 131_072,
+            max_output_tokens: 8_192,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec!["ark-code".into()],
+        },
+        ModelCatalogEntry {
+            id: "doubao-seed-2.0-code".into(),
+            display_name: "Doubao Seed 2.0 Code (Coding Plan)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 262_144,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "doubao-seed-2.0-pro".into(),
+            display_name: "Doubao Seed 2.0 Pro".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Frontier,
+            context_window: 262_144,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "doubao-seed-2.0-lite".into(),
+            display_name: "Doubao Seed 2.0 Lite (Ark Coding)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Fast,
+            context_window: 262_144,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "doubao-seed-code-ark".into(),
+            display_name: "Doubao Seed Code (Ark)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 262_144,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            aliases: vec![],
+        },
+        // Third-party models available via Ark marketplace.
+        // The "ark/" prefix is the canonical ID for these Ark marketplace models to avoid
+        // collisions with native provider entries (minimax, zhipu, moonshot).
+        // The bare model name is kept as an alias only where no collision exists.
+        // Pricing not publicly documented for Ark-routed third-party models; set to 0.0
+        ModelCatalogEntry {
+            id: "ark/minimax-m2.5".into(),
+            display_name: "MiniMax M2.5 (via Ark)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 200_000,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            // "minimax-m2.5" NOT added as alias — already canonical on the minimax provider entry
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "ark/glm-4.7".into(),
+            display_name: "GLM 4.7 (via Ark)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Balanced,
+            context_window: 200_000,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            // "glm-4.7" NOT added as alias — already canonical on the zhipu provider entry
+            aliases: vec![],
+        },
+        ModelCatalogEntry {
+            id: "ark/deepseek-v3.2".into(),
+            display_name: "DeepSeek V3.2 (via Ark)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 131_072,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            // "deepseek-v3.2" kept as alias — no collision with other providers
+            aliases: vec!["deepseek-v3.2".into()],
+        },
+        ModelCatalogEntry {
+            id: "ark/kimi-k2.5".into(),
+            display_name: "Kimi K2.5 (via Ark)".into(),
+            provider: "volcengine_coding".into(),
+            tier: ModelTier::Smart,
+            context_window: 262_144,
+            max_output_tokens: 16_384,
+            input_cost_per_m: 0.0,
+            output_cost_per_m: 0.0,
+            supports_tools: true,
+            supports_vision: false,
+            supports_streaming: true,
+            // "kimi-k2.5" NOT added as alias — already canonical on the moonshot provider entry
+            aliases: vec![],
+        },
+
         // ══════════════════════════════════════════════════════════════
         // Volcano Engine / Doubao (4)
         // ══════════════════════════════════════════════════════════════
+        //
+        // NOTE on separators: the volcengine provider uses hyphen-only IDs
+        // (e.g. "doubao-seed-2-0-lite") because the Ark /api/v3 endpoint uses
+        // endpoint-access-point names that don't contain dots. The
+        // volcengine_coding entries above use dot notation
+        // (e.g. "doubao-seed-2.0-lite") which is the model version string used
+        // by the /api/coding/v3 endpoint. These are different endpoint paths
+        // and the IDs must not be unified.
         ModelCatalogEntry {
             id: "doubao-seed-1-6-251015".into(),
             display_name: "Doubao Seed 1.6 Pro".into(),
@@ -3477,6 +3783,8 @@ fn builtin_models() -> Vec<ModelCatalogEntry> {
             supports_tools: true,
             supports_vision: false,
             supports_streaming: true,
+            // "doubao" also maps to the volcengine provider in provider_defaults() — intentional dual alias
+            // Also matched as a provider alias in provider_defaults() — keep in sync
             aliases: vec!["doubao".into(), "doubao-pro".into()],
         },
         ModelCatalogEntry {
@@ -3507,9 +3815,10 @@ fn builtin_models() -> Vec<ModelCatalogEntry> {
             supports_streaming: true,
             aliases: vec!["doubao-mini".into()],
         },
+        // Standard plan (/api/v3) variant; see volcengine_coding for the coding-plan endpoint
         ModelCatalogEntry {
             id: "doubao-seed-code".into(),
-            display_name: "Doubao Seed Code".into(),
+            display_name: "Doubao Seed Code (Standard Plan)".into(),
             provider: "volcengine".into(),
             tier: ModelTier::Smart,
             context_window: 131_072,
@@ -3905,7 +4214,7 @@ mod tests {
     #[test]
     fn test_catalog_has_providers() {
         let catalog = ModelCatalog::new();
-        assert_eq!(catalog.list_providers().len(), 41);
+        assert_eq!(catalog.list_providers().len(), 42);
     }
 
     #[test]
@@ -4137,6 +4446,10 @@ mod tests {
         assert!(catalog.get_provider("moonshot").is_some());
         assert!(catalog.get_provider("qianfan").is_some());
         assert!(catalog.get_provider("bedrock").is_some());
+        // Alibaba Coding Plan provider
+        assert!(catalog.get_provider("alibaba_coding_plan").is_some());
+        let provider = catalog.get_provider("alibaba_coding_plan").unwrap();
+        assert_eq!(provider.api_key_env, "ALIBABA_CODING_PLAN_API_KEY");
     }
 
     #[test]
@@ -4176,6 +4489,109 @@ mod tests {
         assert!(abab7.supports_vision);
     }
 
+    #[test]
+    fn test_alibaba_coding_plan_models() {
+        let catalog = ModelCatalog::new();
+
+        // Test all 8 models are present
+        let models = catalog.models_by_provider("alibaba_coding_plan");
+        assert_eq!(models.len(), 8);
+
+        // Test qwen3.5-plus — flagship model with 1M context
+        // Note: supports_vision is false — the Coding Plan endpoint is coding-specialized
+        let qwen35 = catalog
+            .find_model("alibaba-coding-plan/qwen3.5-plus")
+            .unwrap();
+        assert_eq!(qwen35.display_name, "Qwen 3.5 Plus (Alibaba Coding Plan)");
+        assert_eq!(qwen35.tier, ModelTier::Smart);
+        assert_eq!(qwen35.context_window, 1_000_000);
+        assert_eq!(qwen35.max_output_tokens, 65_536);
+        assert!(qwen35.supports_tools);
+        assert!(!qwen35.supports_vision);
+        assert!(qwen35.supports_streaming);
+        // Subscription-based pricing
+        assert!((qwen35.input_cost_per_m).abs() < f64::EPSILON);
+        assert!((qwen35.output_cost_per_m).abs() < f64::EPSILON);
+
+        // Test qwen3-max-2026-01-23 — frontier model
+        let qwen3max = catalog
+            .find_model("alibaba-coding-plan/qwen3-max-2026-01-23")
+            .unwrap();
+        assert_eq!(qwen3max.tier, ModelTier::Frontier);
+        assert_eq!(qwen3max.context_window, 262_144);
+        assert!(!qwen3max.supports_vision);
+        assert!(qwen3max.supports_tools);
+
+        // Test qwen3-coder-plus — coding model with 1M context
+        let qwen3coder = catalog
+            .find_model("alibaba-coding-plan/qwen3-coder-plus")
+            .unwrap();
+        assert_eq!(qwen3coder.context_window, 1_000_000);
+        assert!(!qwen3coder.supports_vision);
+        assert!(qwen3coder.supports_tools);
+
+        // Test qwen3-coder alias — registered in builtin_aliases() which uses
+        // or_insert_with (first-wins), so explicit builtin entries take priority
+        // over any model-level aliases vecs that might also claim the same key.
+        let qwen3coder_alias = catalog.find_model("qwen3-coder").unwrap();
+        assert_eq!(qwen3coder_alias.id, "alibaba-coding-plan/qwen3-coder-plus");
+
+        // Verify qwen3 alias still resolves to qwen provider (not overwritten by alibaba aliases)
+        let qwen3 = catalog.find_model("qwen3").unwrap();
+        assert_eq!(qwen3.provider, "qwen");
+
+        // Test glm-5 — Zhipu model via Coding Plan
+        let glm5 = catalog.find_model("alibaba-coding-plan/glm-5").unwrap();
+        assert_eq!(glm5.display_name, "GLM-5 (Alibaba Coding Plan)");
+        assert_eq!(glm5.tier, ModelTier::Frontier);
+        assert_eq!(glm5.context_window, 202_752);
+        assert_eq!(glm5.max_output_tokens, 32_768);
+        assert!(glm5.supports_tools);
+        assert!(!glm5.supports_vision);
+
+        // Test glm-4.7
+        let glm47 = catalog.find_model("alibaba-coding-plan/glm-4.7").unwrap();
+        assert_eq!(glm47.tier, ModelTier::Smart);
+        assert_eq!(glm47.context_window, 202_752);
+
+        // Test kimi-k2.5 — Moonshot model via Coding Plan (no vision on this endpoint)
+        let kimi = catalog.find_model("alibaba-coding-plan/kimi-k2.5").unwrap();
+        assert_eq!(kimi.display_name, "Kimi K2.5 (Alibaba Coding Plan)");
+        assert_eq!(kimi.context_window, 262_144);
+        assert!(!kimi.supports_vision);
+        assert!(kimi.supports_tools);
+
+        // Test MiniMax-M2.5 via Coding Plan
+        let minimax = catalog
+            .find_model("alibaba-coding-plan/minimax-m2.5")
+            .unwrap();
+        assert_eq!(minimax.display_name, "MiniMax M2.5 (Alibaba Coding Plan)");
+        assert_eq!(minimax.context_window, 204_800);
+        assert_eq!(minimax.max_output_tokens, 32_768);
+        assert!(minimax.supports_tools);
+        assert!(!minimax.supports_vision);
+    }
+
+    #[test]
+    fn test_alibaba_coding_plan_aliases() {
+        let catalog = ModelCatalog::new();
+
+        // Test alibaba-coding-plan alias resolves to qwen3.5-plus
+        let alias1 = catalog.find_model("alibaba-coding-plan").unwrap();
+        assert_eq!(alias1.id, "alibaba-coding-plan/qwen3.5-plus");
+
+        // Test qwen3-coder alias resolves to qwen3-coder-plus
+        assert_eq!(catalog.find_model("qwen3-coder").unwrap().id, "alibaba-coding-plan/qwen3-coder-plus");
+
+        // Regression: "qwen-coder" (no digit 3) must remain distinct — it maps to the direct
+        // Qwen Code model, not the Alibaba Coding Plan endpoint.
+        assert_eq!(catalog.find_model("qwen-coder").unwrap().id, "qwen-code/qwen3-coder");
+
+        // Test case-insensitive alias resolution
+        let alias_lower = catalog.find_model("Alibaba-Coding-Plan").unwrap();
+        assert_eq!(alias_lower.id, "alibaba-coding-plan/qwen3.5-plus");
+    }
+
     #[test]
     fn test_bedrock_models() {
         let catalog = ModelCatalog::new();
@@ -4523,4 +4939,69 @@ mod tests {
         assert_eq!(found.provider, "custom_provider");
         assert_eq!(found.id, "My-Custom-LLM");
     }
+
+    #[test]
+    fn test_ark_alias_resolution() {
+        let catalog = ModelCatalog::new();
+        // ark/ IDs are now canonical — resolve_alias returns the id itself (no alias mapping needed)
+        // deepseek-v3.2 is still an alias pointing to ark/deepseek-v3.2
+        assert_eq!(catalog.resolve_alias("deepseek-v3.2"), Some("ark/deepseek-v3.2"));
+        // find_model via ark/ canonical ID returns the volcengine_coding entry directly.
+        let m25 = catalog.find_model("ark/minimax-m2.5").unwrap();
+        assert_eq!(m25.id, "ark/minimax-m2.5");
+        assert_eq!(m25.provider, "volcengine_coding");
+        // glm-4.7 canonical ID now belongs to ark entry; zhipu entry is unaffected.
+        let glm = catalog.find_model("ark/glm-4.7").unwrap();
+        assert_eq!(glm.id, "ark/glm-4.7");
+        assert_eq!(glm.provider, "volcengine_coding");
+        // deepseek-v3.2 exists only under volcengine_coding; bare alias still resolves.
+        let ds = catalog.find_model("ark/deepseek-v3.2").unwrap();
+        assert_eq!(ds.id, "ark/deepseek-v3.2");
+        assert_eq!(ds.provider, "volcengine_coding");
+        let ds_alias = catalog.find_model("deepseek-v3.2").unwrap();
+        assert_eq!(ds_alias.id, "ark/deepseek-v3.2");
+        let kimi = catalog.find_model("ark/kimi-k2.5").unwrap();
+        assert_eq!(kimi.id, "ark/kimi-k2.5");
+        assert_eq!(kimi.provider, "volcengine_coding");
+        // Native provider entries are unaffected by the ark/ rename
+        let minimax_native = catalog.find_model("minimax-m2.5").unwrap();
+        assert_eq!(minimax_native.provider, "minimax");
+        let glm_native = catalog.find_model("glm-4.7").unwrap();
+        assert_eq!(glm_native.provider, "zhipu");
+        let kimi_native = catalog.find_model("kimi-k2.5").unwrap();
+        assert_eq!(kimi_native.provider, "moonshot");
+    }
+
+    #[test]
+    fn test_doubao_alias_resolves_to_volcengine_model() {
+        let catalog = ModelCatalog::new();
+        // "doubao" alias should resolve to the model ID
+        let resolved = catalog.resolve_alias("doubao");
+        assert_eq!(resolved, Some("doubao-seed-1-6-251015"));
+        // The model should belong to the volcengine provider
+        let model = catalog.find_model("doubao-seed-1-6-251015").unwrap();
+        assert_eq!(model.provider, "volcengine");
+    }
+
+    #[test]
+    fn test_doubao_dual_role_consistency() {
+        use openfang_types::model_catalog::VOLCENGINE_BASE_URL;
+
+        // 1. The canonical VOLCENGINE_BASE_URL must point at volces.com
+        assert!(
+            VOLCENGINE_BASE_URL.contains("volces.com"),
+            "VOLCENGINE_BASE_URL must contain volces.com, got: {}",
+            VOLCENGINE_BASE_URL
+        );
+
+        // 2. The model found via the "doubao" alias must belong to the "volcengine" provider
+        let catalog = ModelCatalog::new();
+        let model = catalog
+            .find_model("doubao")
+            .expect("catalog.find_model(\"doubao\") must resolve via alias");
+        assert_eq!(
+            model.provider, "volcengine",
+            "model resolved by 'doubao' must have provider == 'volcengine'"
+        );
+    }
 }
diff --git a/crates/openfang-types/src/model_catalog.rs b/crates/openfang-types/src/model_catalog.rs
index a7d2627ca..a837a3357 100644
--- a/crates/openfang-types/src/model_catalog.rs
+++ b/crates/openfang-types/src/model_catalog.rs
@@ -47,8 +47,19 @@ pub const ZAI_CODING_BASE_URL: &str = "https://api.z.ai/api/coding/paas/v4";
 pub const MOONSHOT_BASE_URL: &str = "https://api.moonshot.ai/v1";
 pub const KIMI_CODING_BASE_URL: &str = "https://api.kimi.com/coding";
 pub const QIANFAN_BASE_URL: &str = "https://qianfan.baidubce.com/v2";
+// Hardcoded to cn-beijing region. Operators in other regions can override via
+// `base_url` in the provider config (e.g. `[provider_urls] volcengine = "https://ark.<region>.volces.com/api/v3"`).
 pub const VOLCENGINE_BASE_URL: &str = "https://ark.cn-beijing.volces.com/api/v3";
 pub const VOLCENGINE_CODING_BASE_URL: &str = "https://ark.cn-beijing.volces.com/api/coding/v3";
+// Alibaba Cloud Coding Plan International — subscription-based
+// See: https://www.alibabacloud.com/help/en/model-studio/coding-plan
+pub const ALIBABA_CODING_PLAN_BASE_URL: &str = "https://coding-intl.dashscope.aliyuncs.com/v1";
+/// Canonical provider ID for Volcano Engine (standard Ark /api/v3 endpoint).
+pub const VOLCENGINE_PROVIDER_ID: &str = "volcengine";
+/// Canonical provider ID for Volcano Engine Coding Plan (Ark /api/coding/v3 endpoint).
+pub const VOLCENGINE_CODING_PROVIDER_ID: &str = "volcengine_coding";
+/// Provider alias for Doubao — routes to the Volcano Engine standard Ark endpoint.
+pub const DOUBAO_PROVIDER_ID: &str = "doubao";
 
 // ── Chutes.ai ────────────────────────────────────────────────────
 pub const CHUTES_BASE_URL: &str = "https://llm.chutes.ai/v1";
diff --git a/docs/providers.md b/docs/providers.md
index 9b3ccb59f..e6276639a 100644
--- a/docs/providers.md
+++ b/docs/providers.md
@@ -1,6 +1,6 @@
 # LLM Providers Guide
 
-OpenFang ships with a comprehensive model catalog covering **3 native LLM drivers**, **20 providers**, **51 builtin models**, and **23 aliases**. Every provider uses one of three battle-tested drivers: the native **Anthropic** driver, the native **Gemini** driver, or the universal **OpenAI-compatible** driver. This guide is the single source of truth for configuring, selecting, and managing LLM providers in OpenFang.
+OpenFang ships with a comprehensive model catalog covering **3 native LLM drivers**, **21 providers**, **61 builtin models**, and **25 aliases**. Every provider uses one of three battle-tested drivers: the native **Anthropic** driver, the native **Gemini** driver, or the universal **OpenAI-compatible** driver. This guide is the single source of truth for configuring, selecting, and managing LLM providers in OpenFang.
 
 ---
 
@@ -549,9 +549,47 @@ For Gemini specifically, either `GEMINI_API_KEY` or `GOOGLE_API_KEY` will work.
 
 ---
 
+### 21. Alibaba Coding Plan (Intl)
+
+| | |
+|---|---|
+| **Display Name** | Alibaba Coding Plan (Intl) |
+| **Driver** | OpenAI-compatible |
+| **Env Var** | `ALIBABA_CODING_PLAN_API_KEY` |
+| **Base URL** | `https://coding-intl.dashscope.aliyuncs.com/v1` |
+| **Key Required** | Yes |
+| **Free Tier** | No (subscription-based) |
+| **Auth** | `Authorization: Bearer` header |
+| **Models** | 8 |
+
+**Available Models:**
+- `alibaba-coding-plan/qwen3.5-plus` (Smart) — 1M context
+- `alibaba-coding-plan/qwen3-max-2026-01-23` (Frontier) — 262K context
+- `alibaba-coding-plan/qwen3-coder-plus` (Smart) — 1M context, coding optimized
+- `alibaba-coding-plan/qwen3-coder-next` (Frontier) — 262K context
+- `alibaba-coding-plan/glm-5` (Frontier) — Zhipu GLM-5, 202K context
+- `alibaba-coding-plan/glm-4.7` (Smart) — Zhipu GLM-4.7, 202K context
+- `alibaba-coding-plan/kimi-k2.5` (Smart) — Moonshot Kimi, 262K context
+- `alibaba-coding-plan/minimax-m2.5` (Smart) — 204K context
+
+**Setup:**
+1. Go to [Alibaba Cloud Model Studio Coding Plan](https://modelstudio.console.alibabacloud.com/ap-southeast-1/?tab=globalset#/efm/coding_plan)
+2. Subscribe to the Pro plan ($50/month)
+3. Copy your plan-specific API key (format: `sk-sp-xxxxx`)
+4. `export ALIBABA_CODING_PLAN_API_KEY="sk-sp-..."`
+
+**Important Notes:**
+- **Subscription-based pricing**: $50/month Pro plan with quota limits (90,000 requests/month, 45,000/week, 6,000 per 5 hours). Per-token costs are $0 — actual cost is the fixed monthly fee.
+- **API Key format**: Must use plan-specific key starting with `sk-sp-`. Regular Model Studio keys (`sk-xxxxx`) will not work.
+- **Base URL**: Must use the Coding Plan endpoint (`coding-intl.dashscope.aliyuncs.com`). The general Model Studio URL will fail.
+- **Multi-provider access**: Single subscription provides access to models from Qwen (Alibaba), GLM (Zhipu), Kimi (Moonshot), and MiniMax.
+- **Vision**: No vision support — this is a coding-specialized endpoint.
+
+---
+
 ## Model Catalog
 
-The complete catalog of all 51 builtin models, sorted by provider. Pricing is per million tokens.
+The complete catalog of all 61 builtin models, sorted by provider. Pricing is per million tokens.
 
 | # | Model ID | Display Name | Provider | Tier | Context Window | Max Output | Input $/M | Output $/M | Tools | Vision |
 |---|----------|-------------|----------|------|---------------|------------|-----------|------------|-------|--------|
@@ -608,6 +646,16 @@ The complete catalog of all 51 builtin models, sorted by provider. Pricing is pe
 | 51 | `grok-2-mini` | Grok 2 Mini | xai | Fast | 131,072 | 32,768 | $0.30 | $0.50 | Yes | No |
 | 52 | `hf/meta-llama/Llama-3.3-70B-Instruct` | Llama 3.3 70B (HF) | huggingface | Balanced | 128,000 | 4,096 | $0.30 | $0.30 | No | No |
 | 53 | `replicate/meta-llama-3.3-70b-instruct` | Llama 3.3 70B (Replicate) | replicate | Balanced | 128,000 | 4,096 | $0.40 | $0.40 | No | No |
+| 54 | `alibaba-coding-plan/qwen3.5-plus` | Qwen 3.5 Plus (Coding Plan) | alibaba_coding_plan | Smart | 1,000,000 | 65,536 | $0.00 | $0.00 | Yes | No |
+| 55 | `alibaba-coding-plan/qwen3-max-2026-01-23` | Qwen 3 Max 2026-01-23 (Coding Plan) | alibaba_coding_plan | Frontier | 262,144 | 65,536 | $0.00 | $0.00 | Yes | No |
+| 56 | `alibaba-coding-plan/qwen3-coder-plus` | Qwen 3 Coder Plus (Coding Plan) | alibaba_coding_plan | Smart | 1,000,000 | 65,536 | $0.00 | $0.00 | Yes | No |
+| 57 | `alibaba-coding-plan/qwen3-coder-next` | Qwen 3 Coder Next (Coding Plan) | alibaba_coding_plan | Frontier | 262,144 | 65,536 | $0.00 | $0.00 | Yes | No |
+| 58 | `alibaba-coding-plan/glm-5` | GLM-5 (Coding Plan) | alibaba_coding_plan | Frontier | 202,752 | 32,768 | $0.00 | $0.00 | Yes | No |
+| 59 | `alibaba-coding-plan/glm-4.7` | GLM-4.7 (Coding Plan) | alibaba_coding_plan | Smart | 202,752 | 32,768 | $0.00 | $0.00 | Yes | No |
+| 60 | `alibaba-coding-plan/kimi-k2.5` | Kimi K2.5 (Coding Plan) | alibaba_coding_plan | Smart | 262,144 | 32,768 | $0.00 | $0.00 | Yes | No |
+| 61 | `alibaba-coding-plan/minimax-m2.5` | MiniMax M2.5 (Coding Plan) | alibaba_coding_plan | Smart | 204,800 | 32,768 | $0.00 | $0.00 | Yes | No |
+
+\* Provider ID uses underscores; model IDs use hyphens. Both forms are accepted by the driver.
 
 **Model Tiers:**
 
@@ -621,13 +669,13 @@ The complete catalog of all 51 builtin models, sorted by provider. Pricing is pe
 
 **Notes:**
 - Local providers (Ollama, vLLM, LM Studio) auto-discover models at runtime. Any model you download and serve will be merged into the catalog with `Local` tier and zero cost.
-- The 46 entries above are the builtin models. The total of 51 referenced in the catalog includes runtime auto-discovered models that vary per installation.
+- The 61 entries above are the builtin models. The total may vary per installation due to runtime auto-discovered models from local providers.
 
 ---
 
 ## Model Aliases
 
-All 23 aliases resolve to canonical model IDs. Aliases are case-insensitive.
+All 25 aliases resolve to canonical model IDs. Aliases are case-insensitive.
 
 | Alias | Resolves To |
 |-------|------------|
@@ -654,6 +702,8 @@ All 23 aliases resolve to canonical model IDs. Aliases are case-insensitive.
 | `sonar` | `sonar-pro` |
 | `jamba` | `jamba-1.5-large` |
 | `command-r` | `command-r-plus` |
+| `alibaba-coding-plan` | `alibaba-coding-plan/qwen3.5-plus` |
+| `qwen3-coder` | `alibaba-coding-plan/qwen3-coder-plus` |
 
 You can use aliases anywhere a model ID is accepted: in config files, REST API calls, chat commands, and the model routing configuration.
 
@@ -778,6 +828,7 @@ The `MeteringEngine` first checks the **model catalog** for exact pricing. If th
 | `*replicate*` | $0.40 | $0.40 |
 | `*llama*` / `*mixtral*` | $0.05 | $0.10 |
 | `*qwen*` | $0.20 | $0.60 |
+| `*alibaba-coding-plan*` | $0.00 | $0.00 **(subscription)** |
 | `mistral-large*` | $2.00 | $6.00 |
 | `*mistral*` (other) | $0.10 | $0.30 |
 | `command-r-plus` | $2.50 | $10.00 |
@@ -903,7 +954,7 @@ Returns a map of all alias-to-canonical-ID mappings.
 GET /api/providers
 ```
 
-Returns all 20 providers with auth status and model counts.
+Returns all 21 providers with auth status and model counts.
 
 **Response:**
 ```json
@@ -998,7 +1049,7 @@ Local:
 
 ### `/providers`
 
-Lists all 20 providers with their authentication status.
+Lists all 21 providers with their authentication status.
 
 ```
 /providers
@@ -1006,7 +1057,7 @@ Lists all 20 providers with their authentication status.
 
 Example output:
 ```
-LLM Providers (20):
+LLM Providers (21):
 
   Anthropic          ANTHROPIC_API_KEY       Configured    3 models
   OpenAI             OPENAI_API_KEY          Missing       6 models
@@ -1047,6 +1098,7 @@ Quick reference for all provider environment variables:
 | Hugging Face | `HF_API_KEY` | Yes |
 | xAI | `XAI_API_KEY` | Yes |
 | Replicate | `REPLICATE_API_TOKEN` | Yes |
+| Alibaba Coding Plan (Intl) | `ALIBABA_CODING_PLAN_API_KEY` | Yes |
 
 ---