Mistral AIFrance· released Oct 2026· open weights

Mistral Large 4

Name: Mistral Large 4
Author: Mistral AI

Ranked #56, tested so far in 4 of the 5 categories the Index averages.

Mistral AI, France. One of 3 models from this lab on the Index.

Third-party publishedHigh confidence (12+ comparable tests, 4+ Index categories)

Snapshot November 10, 2026

OpenCharts Index

45.0

Ranked #56 (52 to 62) of 80 · 55.0% behind the frontier on average · 18 comparable tests · 4 counting categories

Part of this Index is an estimate. On the 4 categories it is tested in, it averages 46.4. The rest is filled in from models tested in every category, because it has too few results to count there yet: Agentic. Such estimates have been within 2.9 points about nine times in ten.

Its band runs from 41.0 to 48.2: where the Index could land once it is tested in every category, or if any single test were dropped. That covers #52 to #62, and neighbours inside that range are ties.

Needs more results (2 of 5 so far)

Averaging 44.2 on the 2 it has; the score opens at 4 of 5.

Distance to today's frontier on the hardest tests. Not a measure of intelligence.

What it's known for

Mistral Large 4, in 18 results.

Strongest in coding (53.3) and math (49.5). It trails the frontier most in multimodal, 60.4% behind on average. Between 50% and 60% behind the best published results, on average. Part of its Index is an estimate. It has too few results in agentic for that category to count, so it is filled in from models tested in every category. On the categories that do count it averages 46.4. Across its band it could sit anywhere from #52 to #62.

Strongest category

Coding

Writing, fixing and shipping real code, judged by tests or users.

53.3

SciCode54.2% · 81% of best
FrontierCode39.4% · 72% of best
Code Arena (WebDev)1,562 · 37% win rate vs leader
WeirdML v312.2% · 23% of best

Mean of 4 of the category's 6 comparable tests. Counts toward the Index.

Best single result

Mock AIME 2024–2025

Solve competition mathematics problems that have a single exact integer answer.

99.4%

99.4% of the best (100.0%) · #15 of 98.

Olympiad-style problems need long, careful chains of reasoning, and there is no partial credit.

Where it trails

Furniture Assembly

Look at a photo of furniture partway through assembly, compare it with the manual, and say whether it has been built correctly so far.

30.8%

98.6% behind the best published result (90.0%), measured above chance · #28 of 37.

Matching a diagram to a real object is the skill behind repairing an appliance or checking an installation from a photo.

Not measured yet

Covered, but not averaged into the Index:

Knowledgesingle-test category, shown beside the Index · 23.8
Agentic2 of 6 tests · needs 3 · mean 40.7
Human preferencesingle-test category, shown beside the Index · 71.2

Every score, with its source

See it at work

What Mistral Large 4 was asked to do, and how it did.

Every test it has taken, by category, with the kind of task it faced, who produced the number and where the result landed against the best published one. The examples are original and illustrative, never items from the datasets themselves.

Human preference · mean 71.2 across 1 comparable test · a single-test category, shown beside the Index and never averaged into it

Text ArenaLeaderboard

Answer the same prompt as a rival model; real users vote blind on which answer they prefer.

1,423

36% expected win rate against the board leader (1,526)
#91 of 113

board 2026-11-09

Source: Arena

Every result is read against the best published one on its test: percent scores as a share of the best above chance, Arena ratings as the expected win rate against the board leader, open-ended values on a log scale. Sitting a harder exam never lowers a model. Every number links to the publisher that produced it.

Head to head

Mistral Large 4 against whoever you pick.

Choose a rival. Every counted test both have taken appears side by side as a share of the best published result on that test, with the category means above. A win is a higher share; a gap under 2 points is a tie, because one item on a 45-problem exam is 2.2 points and publishers' own standard errors are of that order.

Mistral Large 4

#56·Index 45.0D

0 : 17

wins · 1 tie · 18 shared

GPT-6 Astra

#1·Index 93.7A

GPT-6 Astra comes out ahead on 17 of the 18 tests both have taken.

Category means

Reasoning
43.4
96.1
Knowledge
23.8
80.4
Coding
53.3
93.4
Math
49.5
98.8
Agentic
40.7
90.6
Multimodal
39.6
89.6
Human preference
71.2
85.2

Shared tests · biggest gaps first

ProofBenchMath
10.0%
99.0%
−89.0
Furniture AssemblyMultimodal
30.8%
80.0%
−81.9
Mystery Game PuzzlesReasoning
20.0%
84.0%
−79.3
WeirdML v3Coding
12.2%
54.0%
−77.4
FrontierMath Tier 4Math
26.8%
97.6%
−70.8
EBR-benchAgentic
18.1%
76.2%
−64.2
CritPtReasoning
11.1%
31.7%
−57.1
SimpleQA VerifiedKnowledge
22.4%
75.6%
−56.6
Chess PuzzlesReasoning
33.0%
74.0%
−55.3
Code Arena (WebDev)Coding
1,562
1,767
−48.1
FrontierMath (Tiers 1–3)Math
58.6%
93.7%
−36.9
FrontierCodeCoding
39.4%
53.3%
−25.3
APEX-AgentsAgentic
51.6%
64.7%
−15.5
Vision ArenaMultimodal
1,231
1,283
−14.8
Text ArenaHuman preference
1,423
1,474
−14.0
GPQA DiamondReasoning
90.2%
95.8%
−7.9
SciCodeCoding
54.2%
56.5%
−3.4
Mock AIME 2024–2025Math
99.4%
100.0%
tie

Bars are distances to the best published result on each test (percent scores above chance, Arena ratings as win rate against the leader, open-ended values on a log scale). Raw values as the publishers report them.

Open GPT-6 Astra

Scores

Every test, every source

The best published run per benchmark, the raw value as the publisher reports it, who produced the number, how close it comes to the best published result on that test, and where the model sits among every model scored on it. Percent scores are read above chance, Arena ratings as a win rate against the board leader, open-ended values on a log scale. A test with too few models to compare against is shown but not counted, and so is an archived test: one whose publisher no longer covers new models.

Category means

Reasoning43.4 · 4/8
Knowledgebeside23.8 · 1/1
Coding53.3 · 4/6
Math49.5 · 4/4
Agentic40.7 · 2/6
Multimodal39.6 · 2/4
Human preferencebeside71.2 · 1/1
Long context—

Mean · tests taken of the category's comparable tests. A faded category either rests on fewer than half of them or is a single-test category (marked “beside”), which is shown next to the Index and never averaged into it.

GPQA DiamondEpoch-runReasoning

90.2%

91.2% of the best (96.4%), above chance
55th percentile of 96 · n = 198

high

Mistral Large 4

Mistral Large 4, in 18 results.

What Mistral Large 4 was asked to do, and how it did.

Mistral Large 4 against whoever you pick.

Every test, every source

Compare with its neighbours.

The best models, ranked here, working inside Theo.