Ranked by blended price: (input + 3× output) ÷ 4, the industry-standard weighting that reflects realistic chat-style traffic. Lower is cheaper.
Data updated: August 11, 2026
| # | Model | Vendor | Arena Elo | SWE-bench | Price in/out ($/M) | Context |
|---|
The table above sorts by blended price per million tokens. Small models like Gemini Flash, GPT-5 mini and Claude Haiku are one to two orders of magnitude cheaper than the flagships and handle most high-volume production workloads.
Multiply requests per month by the tokens each request consumes and by the price per million tokens. Our free LLM cost calculator does it for 40+ models at once with official list prices.