AI Model Pricing Reference
What this is: one page per current model with verified rates and worked workload costs, plus head-to-head comparisons that price the same task twice. Every rate links to the provider’s own documentation and carries its check date. Last full verification: September 18, 2026.
Model rate cards
Anthropic
- Claude Fable 5.1 — $10.00 / $50.00 per 1M, cache reads $0.25, 1M context. A 200-step agent task bills about $7.49.
- Claude Opus 5 — $5.00 / $25.00 per 1M, cache reads $0.50, 1M context. A 200-step agent task bills about $6.73.
- Claude Sonnet 5 — $2.00 / $10.00 per 1M, cache reads $0.20, 1M context. A 200-step agent task bills about $2.69.
- Claude Haiku 4.5 — $1.00 / $5.00 per 1M, cache reads $0.10, 200K context. A 200-step agent task bills about $1.35.
OpenAI
- GPT-6 Astra — $10.00 / $50.00 per 1M, cache reads $1.00, 1.05M context. A 200-step agent task bills about $13.46.
- GPT-5.6 Sol — $4.00 / $20.00 per 1M, cache reads $0.40, 1.05M context. A 200-step agent task bills about $5.38.
- GPT-5.6 Terra — $2.00 / $12.00 per 1M, cache reads $0.20, 1.05M context. A 200-step agent task bills about $2.89.
- GPT-5.6 Luna — $0.20 / $1.20 per 1M, cache reads $0.020, 1.05M context. A 200-step agent task bills about $0.29.
Google DeepMind
- Gemini 3.1 Pro — $2.00 / $12.00 per 1M, cache reads $0.20, 1M context. A 200-step agent task bills about $2.79.
- Gemini 3.8 Flash — $0.75 / $3.75 per 1M, cache reads $0.075, 1M context. A 200-step agent task bills about $0.97.
DeepSeek
- DeepSeek V4.1 Flash — $0.15 / $0.60 per 1M, cache reads $0.0030, 1M context. A 200-step agent task bills about $0.09.
- DeepSeek V4 Pro — $0.66 / $1.98 per 1M, cache reads $0.022, 1M context. A 200-step agent task bills about $0.40.
Head-to-head comparisons
- Claude Fable 5.1 vs GPT-6 Astra — $7.49 vs $13.46 on the same 200-step task.
- Claude Sonnet 5 vs GPT-5.6 Sol — $2.69 vs $5.38 on the same 200-step task.
- Claude Opus 5 vs GPT-5.6 Sol — $6.73 vs $5.38 on the same 200-step task.
- Gemini 3.1 Pro vs GPT-5.6 Terra — $2.79 vs $2.89 on the same 200-step task.
- GPT-5.6 Luna vs Gemini 3.8 Flash — $0.29 vs $0.97 on the same 200-step task.
- GPT-5.6 Luna vs DeepSeek V4.1 Flash — $0.29 vs $0.09 on the same 200-step task.
- Gemini 3.8 Flash vs DeepSeek V4.1 Flash — $0.97 vs $0.09 on the same 200-step task.
- Claude Haiku 4.5 vs Gemini 3.8 Flash — $1.35 vs $0.97 on the same 200-step task.
Method
Rates come from provider documentation, never from third-party comparison tables; the verification date is republished with every figure. Workload assumptions are stated on each page and reproducible in the cost calculator. The verification workflow itself is documented in our methodology, and corrections are logged publicly on the about page.