Pick two or three models and see exactly how far apart they are — price per million tokens, context window, and what each one can actually do.
| METRIC | MiniMax M3 MiniMax | GPT-6 Luna OpenAI | Claude Haiku 4.5 Anthropic |
|---|---|---|---|
| Input $/1M | $0.30+200% | $0.10CHEAPEST | $1.00+900% |
| Cached input $/1M | $0.06+500% | $0.01CHEAPEST | $0.10+900% |
| Output $/1M | $1.20+140% | $0.50CHEAPEST | $5.00+900% |
| Blended $/1M (3:1) | $0.525+162% | $0.20CHEAPEST | $2.00+900% |
| Context window | 1.05M−0% | 1.05MLARGEST | 200K−81% |
| Capabilities | REASONINGVISIONVIDEO INPUTTOOL USEOPEN WEIGHTS | REASONINGVISIONTOOL USE | REASONINGVISIONTOOL USE |
Percentages are relative to the “best” value in each row — cheapest price, largest context. Blended assumes a typical 3:1 input-to-output token ratio.
ReqKey gives every API key a credit balance — validate, deduct, enforce limits, and see per-consumer spend in real time.
Start metering free