Back

Mistral

Mistral Medium 3.1

Closed weightsModels.dev
Intelligence
9.2
Rank 453 of 744 measured models
Coding
20.5
Rank 226 of 298 measured models
Output speed
149 t/s
Rank 87 of 403 measured models
Price/1M
$0.80
Rank 340 of 659 measured models
Context window
128K
Rank 773 of 1199 measured models

Mistral Medium 3.1 scores 9.2 on the Artificial Analysis Intelligence Index, ranking 453rd of 744 measured models. The closest models on this index are Ling-1T (9.2), Motif-2-12.7B-Reasoning (9.2), and Nova Premier (9.2). At $0.80 per million tokens (3:1 input/output blended price), it is cheaper than Nova Premier ($5).

Artificial Analysis Indices
Intelligence9.2
Rank 453 of 744 measured models
Coding20.5
Rank 226 of 298 measured models
Math38.3
Rank 164 of 270 measured models
Agentic1.9
Rank 141 of 189 measured models
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting8.7
Rank 195 of 218 measured models
Strategy & Ops8.1
Rank 177 of 218 measured models
Legal9.7
Rank 198 of 218 measured models
Engineering10.7
Rank 202 of 217 measured models
Economics13.6
Rank 203 of 225 measured models
Standard Benchmarks
MMLU Pro68.3%
Rank 236 of 349 measured models
GPQA Diamond58.8%
Rank 415 of 619 measured models
HLE4.7%
Rank 504 of 664 measured models
LiveCodeBench40.6%
Rank 175 of 343 measured models
SciCode32.4%
Rank 204 of 231 measured models
AIME 202538.3%
Rank 164 of 270 measured models
IFBench39.8%
Rank 278 of 450 measured models
AA-LCR42.7%
Rank 357 of 575 measured models
Additional benchmarks8
Design Arena · 3d44.6%
Design Arena · Asciiart30.8%
Design Arena · Codecategories45.1%
Design Arena · Dataviz47.2%
Design Arena · Gamedev40.8%
Design Arena · Svg38.2%
Design Arena · Uicomponent43.6%
Design Arena · Website46.0%
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy