Skip to content
aiqbench

Mistral · model passport

Mistral Large 3

mistral-large-3
  • Open weights
  • Apache 2.0 licence
  • Released 2025-10-24(curated)
10/27axes covered
Overall · LMArena
1,414#132 of 396
Best percentile
71.1Coding
Cohort standing
52.0median 53.3

What the data says

Benchmarked on 10 of 27 tracked axes. Every figure below is the model's own published result on that board.

Weakest axes

  • 8.6 pct — WebDev Arena (1,230)

Uneven: 62.5 percentile points between its best and worst axis — 71.1 pct on Coding versus 8.6 pct on WebDev Arena.

  • Coding71.1 pct1,471Verify →
  • MMLU-Pro (Artificial Analysis)70.4 pct80.7%Verify →
  • Instruction following67.2 pct1,402Verify →
  • Overall66.9 pct1,414Verify →
  • Hard prompts66.7 pct1,432Verify →
  • LiveCodeBench (Artificial Analysis)54.2 pct46.5%Verify →
  • Vision Arena46 pct1,200Verify →
  • Factuality Arena44.2 pct1,434Verify →
  • IFBench (Artificial Analysis)30.9 pct36.2%Verify →
  • WebDev Arena8.6 pct1,230Verify →
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • GPQA DiamondNot benchmarked
  • AIME (math)Not benchmarked
  • FrontierMathNot benchmarked
  • SimpleQA VerifiedNot benchmarked
  • ARC-AGI-2Not benchmarked
  • Humanity's Last ExamNot benchmarked
  • SimpleBenchNot benchmarked
  • LiveBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • HHEM Hallucination (Vectara)Not benchmarked
  • SWE-bench VerifiedNot benchmarked
  • Terminal-BenchNot benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

  • mistral-large-3 stands at 52 (raw mean 52.6) — measured on 10 of 27 axes.

    standingtemplateVerify →

  • mistral-large-3 stands at 52 versus a cohort median of 53.3 across 137 models released within 180 days — below the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →