chore(pricing): Update dashscope pricing

[siddharthsambharia-portkey]

🔄 Pricing Update: dashscope

📊 Summary (complete_diff mode)

Change Type	Count
➕ Models added	168
🔄 Models updated (merged)	37

➕ New Models

• qwen-plus-us
• qwen-plus-2025-12-01-us
• qwen-flash-latest
• qwen-flash-us
• qwen-flash-2025-07-28-us
• qwen-deep-research
• qwen3-vl-flash-us
• qwen3-vl-flash-2026-01-22-us
• qwen3-vl-flash-2025-10-15-us
• qwen2.5-math-72b-instruct
• qwen2.5-math-7b-instruct
• qwen2.5-math-1.5b-instruct
• qwen-vl-ocr-2025-08-28
• qwen3-235b-a22b-instruct
• qwen3-235b-a22b-thinking
• qwen3-235b-a22b-instruct-2507
• qwen3-235b-a22b-thinking-2507
• qwen3-32b-instruct
• qwen3-30b-a3b-instruct
• qwen3-30b-a3b-thinking
• ... and 148 more

🔄 Updated Models

• qvq-plus
• qvq-plus-latest
• qvq-plus-2025-05-15
• qwen3-coder-next
• qwen-vl-ocr-latest
• qwen3-32b
• qwen3-next-80b-a3b-instruct
• qwq-32b
• qwq-32b-preview
• qwen3-vl-235b-a22b-instruct
• qwen3-vl-235b-a22b-thinking
• qwen3-vl-30b-a3b-thinking
• qwen3-vl-8b-thinking
• qwen2-vl-72b-instruct
• qwen3-omni-flash
• qwen3-omni-flash-2025-12-01
• qwen3-omni-flash-2025-09-15
• qwen-omni-turbo
• qwen-omni-turbo-latest
• qwen-omni-turbo-2025-03-26
• qwen-omni-turbo-2025-01-19
• qwen3-omni-flash-realtime
• qwen-omni-turbo-realtime
• qwen-omni-turbo-realtime-latest
• qwen-omni-turbo-realtime-2025-05-08
• qwen3-omni-30b-a3b-captioner
• qwen3-livetranslate-flash-realtime
• deepseek-v3.2
• glm-4.6
• qwen-plus-character
• ... and 7 more

Model-to-Pricing-Page Mapping

Sources

• Page 1: https://www.alibabacloud.com/help/en/model-studio/models — Commercial Qwen tiers
• Page 2: https://www.alibabacloud.com/help/en/model-studio/model-pricing — All other models (open-source, speech, video, image, embeddings, third-party)

Regional Priority Applied

International → Global → Chinese Mainland (fallback) per model

Notable Pricing Rules

• Tiered pricing: Lowest tier used as default (e.g. qwen3-max 0–32K: $1.2/$6.0)
• Non-thinking rates: Used where both thinking/non-thinking exist for same model ID
• qwen-deep-research: per_thousand_tokens (not per_million) — $0.007742/$0.023367 per 1K tokens
• TTS models: per_million_characters (scaled from per 10K chars on page)
• ASR/Video models: per_second
• Image models: per_image
• Mainland-only models: qwen-long, qvq-plus, qwen-math-, qwen-coder (old), qwen-deep-research, qwen2.5-math-, paraformer-, tongyi-intent-detect-v3, gte-rerank-v2, qwen-doc-turbo, deepseek-, kimi-, MiniMax-, GLM-*

Model Families Covered (~283 models)

• Flagship: qwen3-max, qwen3.5-plus/flash
• Commercial Qwen: qwen-max/plus/flash/turbo + US variants + dated snapshots
• QwQ/QVQ commercial + open-source
• Open-source Qwen3 (0.6b–235b), Qwen3.5 (27b–397b), Qwen2.5 (0.5b–72b)
• Open-source VL: qwen3-vl-/qwen2.5-vl-/qwen2-vl-*
• Omni/multimodal: qwen3-omni-flash, qwen-omni-turbo, qwen3.5-omni-*, realtime variants
• Coder: qwen3-coder-/qwen2.5-coder-/qwen-coder-*
• Math: qwen-math-/qwen2.5-math-
• Translation: qwen-mt-plus/flash/lite/turbo
• OCR: qwen-vl-ocr
• VL commercial: qwen3-vl-plus/flash, qwen-vl-max/plus
• TTS: qwen3-tts-, qwen-tts-, cosyvoice-*
• ASR: qwen3-asr-, fun-asr-, paraformer-*
• LiveTranslate: qwen3-livetranslate-flash-realtime
• Embeddings: text-embedding-v3/v4, tongyi-embedding-vision-*, qwen3-vl-embedding
• Rerank: qwen3-rerank, gte-rerank-v2
• Domain: tongyi-intent-detect-v3, qwen-*-character, qwen-doc-turbo
• Image gen: qwen-image-, wan2.-t2i-*, z-image-turbo
• Video gen: wan2.-t2v-, wan2.-i2v-, wan2.-kf2v-, wan2.-r2v-, wan2.-vace-, digital human models
• Third-party: deepseek-, kimi-, MiniMax-M2.5, glm-*

---

Generated by Pricing Agent on 2026-04-27