Mistral AI, released Oct 6, 2026

Mistral Large 4price, context, benchmarks and release details

Provisional: not enough results to rank yet 64% confidence, 2 of 6 expected sources in

62.2

SI Score

#53 of 142 ranked models

Input, per 1M tokens: $0.68
Output, per 1M tokens: $2.09
Context window: 1M
Max output: 262K
Released: Oct 6, 2026

How the score breaks down

Coding 65.2

Math 88.1

Preference 75.7

Reasoning 77.4

Weights: reasoning 30%, math 15%, coding 40%, preference 15%. Results use fixed 0–100 scales before averaging, and thin evidence is pulled toward 50. Method si-v3-retained-evidence-2, computed Oct 9, 2026, 05:13 UTC.

Around it on the leaderboard

51 MiMo-V2.5-Pro 62.6
52 Muse Spark 1.2 62.3
53 Mistral Large 4 62.2
54 Qwen3.8 27B 61.9
55 GPT-6 Luna 61.6

Full leaderboard

Benchmark results

17 benchmarks, 17 results

Each row shows the best published result. Where a model was tested at several settings, such as reasoning effort, open the row to see each one. Hover or tap a value for its source.

LiveBench Reasoning: connectionsreasoning 48% 48.0

LiveBench Reasoning: consecutive eventsreasoning 80.5% 80.5

LiveBench Reasoning: logic with navigationreasoning 68% 68.0

LiveBench Reasoning: spatialreasoning 94% 94.0

LiveBench Reasoning: theory of mindreasoning 80.8% 80.8

LiveBench Reasoning: zebra puzzlesreasoning 93% 93.0

LiveBench Math: AMPS Hardmath 98% 98.0

LiveBench Math: competition mathmath 97.1% 97.1

LiveBench Math: integralsmath 90% 90.0

LiveBench Math: olympiadmath 89.3% 89.3

LiveBench Math: simplifymath 66.2% 66.2

LiveBench Coding: code completioncoding 78.3% 78.3

LiveBench Coding: code generationcoding 76.1% 76.1

LiveBench Coding: JavaScriptcoding 68.2% 68.2

LiveBench Coding: Pythoncoding 60% 60.0

LiveBench Coding: TypeScriptcoding 43.3% 43.3

LMArena Textpreference 1427.4 elo 75.7

Normalization uses a fixed 0–100 scale for each unit, independent of other models. Compare evaluation conditions before reading a small gap as decisive. “Lab-reported” marks the provider's own published figure.

Details and sources

Open weights: No
License: not yet reported
Input modalities: text, image
First seen by SuperIndex: Oct 8, 2026
Coverage: 51% of expected source weight

Reported (2)

LiveBenchOct 8, 2026
LMArena / ArenaOct 9, 2026

Awaiting (4)

ARC Prize17% of weight
Humanity’s Last Exam17% of weight
Official model cards via models.dev6% of weight
Terminal-Bench9% of weight

Confidence rises as pending sources publish. Some sources never cover some models, so confidence reaches 100% at 80% of expected weight.