MiniMax: MiniMax M2
minimax/minimax-m2
Cheapest provider
$0.26 / 1M
AtlasCloud
Fastest provider (p95)
187 tok/s
Intelligence (composite)
80.5
MMLU-Pro · HumanEval · math · GPQA
Per-provider performance
Latency / throughput / uptime / price measured across providers over the last 30 minutes of live traffic. This is what proves “sourced cheapest” — Atlas mode draws on these per call to serve the cheapest path that holds quality.
| Provider | Quant | Input $/1M | Output $/1M | Latency p50 / p95 | Throughput p50 / p95 | Uptime 30m | Success |
|---|---|---|---|---|---|---|---|
| AtlasCloud | q8· fp8 | $0.2550 | $1.0000 | 1739ms / 8382ms | 7 / 48.8 tok/s | — | 95.5% |
| Minimax | q8· fp8 | $0.2550 | $1.0200 | 1312ms / 4120ms | 10 / 77.5 tok/s | — | 100.0% |
| undisclosed | $0.3000 | $1.2000 | 3044ms / 4399ms | 158 / 187 tok/s | — | 100.0% | |
| Novita | q8· fp8 | $0.3000 | $1.2000 | 1465ms / 8372ms | 75 / 97 tok/s | — | 100.0% |
“—” means live telemetry hasn’t accumulated enough recent traffic for that endpoint. “undisclosed” means the provider serves the model but doesn’t expose the quantization label (typically running fp8 / int8 internally).
Intelligence breakdown
Composite score is a weighted average of public benchmarks (30% MMLU-Pro, 25% code pass@1, 25% math, 20% GPQA). Numbers come from model cards and the Artificial Analysis intelligence harness; missing components are renormalised over what’s present.
MMLU-Pro
82.0
broad reasoning
Code
83.0
pass@1 (HumanEval / LiveCodeBench)
AIME 2025
78.0
math accuracy
GPQA Diamond
78.0
hard reasoning
Source: Artificial Analysis (AA-aligned) — MiniMax-M2 (MMLU-Pro 82, GPQA-D 78, LiveCodeBench 83, AIME25 78); github.com/MiniMax-AI/MiniMax-M2
How Atlas mode sources MiniMax: MiniMax M2
- Strict mode — pin MiniMax: MiniMax M2 exactly and we pass it straight through, sourced from the cheapest provider above. The same model, no substitutions — currently AtlasCloud at $0.26/1M.
- Atlas mode — the default. Each call is auto-optimized for the cheapest path that holds quality, at least 5% off going direct from call one and climbing as it ramps. You always see which model served the call and exactly what you saved — thumbs-down anything you don’t like for a full refund.