MDL
MiniMax-M3
MiniMax
Verified against Artificial Analysis · Jul 13, 2026
Capability
94
percentile index
Price /M
$0.3 / $1.2
in / out
Context
1M
tokens
Avg score
66.2
7 benchmarks
AA Intelligence
44.4
index
Speed
111 tok/s
TTFT 1.4s
Released
Jun 1, 2026
About
MiniMax-M3 is a open-weights model in the MiniMax family from MiniMax. Benchmarked on 7 evals, averaging 66.2.
Benchmark scores · 7
avg 66.2 — higher is betterThe research behind this model
all papers mentioning it →MiniMax Sparse Attention▲ 141 on HF · Jun 11, 2026MaxProof: Scaling Mathematical Proof with Generative-Verifier RL and Population-Level Test-Time Scaling▲ 90 on HF · Jun 11, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed111 tok/s
Latency (TTFT)1.4s
First answer19s
Price in / out$0.3 / $1.2 per 1M
Blended price$0.525 / 1M
Coding index58.6
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$1.2/M