Open-Source wins. You can't lose what you give freely. Knowledge shared, multiplies.
Shipping with AI agents around the clock -- human hours for thinking, machine hours for doing.
Stats auto-updated by aidevops.
| Metric | 24h | 7 Days | 28 Days | 365 Days |
|---|---|---|---|---|
| Screen time (Mac) | 15.4h | 103.9h | 365h | ~4,672h* |
| User AI session hours | 1.8h | 38.6h | 110.0h | 352.4h |
| AI worker hours | 11.3h | 80.4h | 236.2h | 309.9h |
| AI concurrency hours | 14.7h | 201.6h | 519.7h | 1,114.7h |
| Interactive sessions | 8 | 73 | 263 | 1,054 |
| Worker sessions | 165 | 1,234 | 3,565 | 4,030 |
Screen time from macOS display events, snapshotted daily. *365-day extrapolated (accumulating real data).
User AI session hours measured from AI message timestamps (reading, thinking, typing between responses).
| Model | Requests | Input | Output | Cache read | API Cost | Cache savings | Model savings |
|---|---|---|---|---|---|---|---|
| claude-opus-4-6 | 84,835 | 91K | 26.2M | 6,496.5M | $16,134.00 | $87,702.90 | $0.00 |
| gpt-5.3-codex | 23,276 | 132.3M | 6.2M | 1,587.0M | $1,121.28 | $2,975.64 | $3,449.04 |
| claude-sonnet-4-6 | 38,177 | 40K | 10.5M | 2,012.7M | $1,050.80 | $5,434.37 | $3,046.45 |
| gpt-5.4 | 2,677 | 16.6M | 683K | 193.2M | $120.08 | $362.38 | $422.20 |
| claude-opus-4-5 | 10 | 2 | 5K | 377K | $1.95 | $5.10 | $0.00 |
| big-pickle | 69 | 81K | 14K | 2.9M | $1.83 | $0.00 | $6.72 |
| claude-haiku-4-5 | 135 | 280 | 17K | 1.8M | $0.60 | $1.32 | $3.85 |
| minimax-m2.5-free | 12 | 10K | 1K | 211K | $0.15 | $0.00 | $0.61 |
| Total | 149,191 | 149.2M | 43.7M | 10,294.8M | $18,430.69 | $96,481.70 | $6,928.88 |
10,832.4M total tokens processed. 95% cache hit rate.
$103,410.58 total saved ($96,481.70 caching + $6,928.88 model routing vs all-Opus).
Model savings are modest because ~95% of tokens are cache reads, where price differences between models are small.
| Model | Requests | Input | Output | Cache read | API Cost | Cache savings | Model savings |
|---|---|---|---|---|---|---|---|
| claude-opus-4-6 | 134,128 | 143K | 41.1M | 10,854.1M | $19,372.83 | $146,531.08 | $0.00 |
| claude-opus-4-5 | 42,226 | 18K | 11.9M | 3,731.6M | $6,497.37 | $50,377.03 | $0.00 |
| gpt-5.3-codex | 23,667 | 132.4M | 6.2M | 1,587.5M | $1,385.71 | $2,976.63 | $3,450.32 |
| claude-sonnet-4-6 | 38,875 | 40K | 10.4M | 2,008.3M | $759.45 | $5,422.47 | $3,037.85 |
| claude-sonnet-4-5 | 8,161 | 13K | 8.0M | 452.1M | $256.11 | $1,220.83 | $1,024.49 |
| gpt-5.4 | 2,714 | 16.6M | 683K | 193.2M | $169.35 | $362.38 | $422.20 |
| gpt-5.2-codex | 798 | 6.7M | 249K | 54.6M | $53.53 | $102.42 | $148.56 |
| gpt-5.1-codex-max | 62 | 735K | 11K | 2.8M | $3.73 | $5.39 | $12.48 |
| gpt-5.2 | 69 | 300K | 35K | 4.0M | $3.60 | $7.51 | $9.55 |
| gemini-3-pro | 81 | 923K | 10K | 6.5M | $3.30 | $6.18 | $21.18 |
| antigravity-claude-sonnet-4-5 | 51 | 322K | 17K | 3.6M | $2.31 | $9.89 | $9.34 |
| gemini-3-pro-preview | 311 | 543K | 12K | 4.6M | $2.25 | $4.39 | $13.85 |
| claude-sonnet-4 | 101 | 179 | 9K | 1.9M | $0.73 | $5.34 | $2.97 |
| claude-haiku-4-5 | 207 | 519 | 55K | 5.2M | $0.64 | $3.81 | $11.46 |
| grok-code | 2 | 156K | 378 | 256 | $0.47 | $0.00 | $1.91 |
| antigravity-gemini-3-pro | 24 | 101K | 2K | 616K | $0.33 | $0.58 | $2.31 |
| claude-3-haiku | 31 | 106 | 5K | 694K | $0.28 | $1.88 | $1.17 |
| Total | 251,508 | 159.1M | 79.0M | 18,912.1M | $28,511.99 | $207,037.82 | $8,169.64 |
19,875.1M total tokens processed. 95.2% cache hit rate.
$215,207.45 total saved ($207,037.82 caching + $8,169.64 model routing vs all-Opus).
Model savings are modest because ~95.2% of tokens are cache reads, where price differences between models are small.
- homebrew-tap -- Homebrew tap for aidevops and other tools
- aidevops -- Vibe-Coding is easy. DevOps is hard. AI DevOps automates your software, business, and personal development with managed infrastructure through AI chat in OpenCode. Opinionated tools, services, CLI & API tech-stack — for speed, security, and 24/7 results. Open-source-preferred, and SOTA everything.
- aidevops.sh -- The AI Devops website source code for https://aidevops.sh
- quickfile-mcp -- MCP server for QuickFile UK accounting software - invoices, clients, purchases, banking, and reports
- cloudron-netbird-app -- NetBird WireGuard mesh VPN - Cloudron app package
- aidevops-cloudron-app -- Cloudron app: always-on remote worker node for AI DevOps — headless Claude Code dispatch, multi-node orchestration
- amazon-order-history-csv-download-mcp -- Download Amazon Order History As Orders & Items CSV files, plus simplified CSV file versions for importing to your accounting software. Supports Amazon United States US, United Kingdom, Canada, Germany, France, Spain, Italy, Netherlands, Japan, Australia, Mexico, India, UAE, Saudi Arabia, Ireland, Belgium
- marcusquinn.github.io -- Personal Website, Work, Ecommerce, Social Links, Software, Hosting, Services, Preferences, Audio, Music, Video, Markdown, Plain Text, Links
- bash-language-server -- A language server for Bash
- ultimate-multisite -- Website as a service plugin for WordPress Multisite
- claude-code-mcp -- Claude Code as one-shot MCP server to have an agent in your agent.
- opencode -- The open source coding agent.
- aidevops-dashboard -- AI DevOps Dashboard
- opencode-1 -- A powerful AI coding agent. Built for the terminal.
- awesome-selfhosted -- A list of Free Software network services and web applications which can be hosted locally. Selfhosting is the process of hosting and managing applications instead of renting from Software-as-a-Service providers
- wp-performance-action -- Reusable GitHub action to measure performance metrics of WordPress sites
- Turing-CSS -- Typora theme inspired by the style of typewritten articles
Stats auto-updated 2026-03-16 02:20 UTC by aidevops pulse.