MMODELYST
Models/Mistral Medium 3
MDL

Mistral Medium 3

Mistral

Verified against Artificial Analysis · Jul 13, 2026
Capability
52
percentile index
Price /M
$0.4 / $2
in / out
Context
131.1K
tokens
Avg score
38.9
11 benchmarks
AA Intelligence
12.5
index
Speed
47 tok/s
TTFT 0.5s
Released
May 7, 2025
About

Mistral Medium 3 is a open-weights model in the Mistral family from Mistral. Benchmarked on 11 evals, averaging 38.9.

Benchmark scores · 11
avg 38.9 — higher is better
MATH
90.7
MMLU-Pro
76
GPQA Diamond
57.8
LiveCodeBench
40
IFBench
39.3
SciCode
33.1
AIME
30.3
AA Long-Context Reasoning
28
τ²-bench
24.3
Humanity's Last Exam
4.3
Terminal-Bench Hard
3.8
The research behind this model
all papers mentioning it →

Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.

Details
PublisherMistral
FamilyMistral
Parameters
License
WeightsOpen
ReleasedMay 7, 2025
Performance & price
Output speed47 tok/s
Latency (TTFT)0.5s
Price in / out$0.4 / $2 per 1M
Blended price$0.8 / 1M
Math index30.3

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers
Artificial Analysis$2/M