BenchLeader

Mistral Large 4 vs Mistral Large 4.0 OSS

Full profiles: Mistral Large 4 (Mistral AI) and Mistral Large 4.0 OSS (Mistral AI): every benchmark, price and rank.

Updated · Data from 7 sources

As of 16 Nov 2026, there are not enough shared benchmarks for a verdict. Choose Large 4 for knowledge (41 vs 34); choose Large 4.0 OSS for reasoning (64 vs 46) and maths (66 vs 54).

Key facts (as of 16 Nov 2026)

  • Mistral Large 4 costs $1.05 per 1M tokens blended (list price, 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
MathsMistral Large 4.0 OSS66 vs 54 category scoreHigh
ReasoningMistral Large 4.0 OSS64 vs 46 category scoreHigh
KnowledgeMistral Large 441 vs 34 category scoreMedium
Chat quality (human preference)–Only Mistral Large 4 has an Arena rating (60 category score)Not comparable
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • Not enough shared benchmarks for a verdict.
  • Mistral Large 4 is stronger in knowledge.
  • Mistral Large 4.0 OSS (high) is stronger in maths, reasoning.

At a glance

SpecMistral Large 4Mistral Large 4.0 OSS
DeveloperMistral AIMistral AI
Released6 Oct 202629 Oct 2026
Context window (tokens)1,048,576–
Compared atMistral Large 4Mistral Large 4.0 OSS (high)
Open weightsYesNo
Price in / out / cached input per 1M$0.700 / $2.10 / $0.070–

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryMistral Large 4Mistral Large 4.0 OSS (high)Leader
Reasoning46.264.2Mistral Large 4.0 OSS
Coding53.8–not comparable
Agents & tools48.8–not comparable
Maths54.266.4Mistral Large 4.0 OSS
Knowledge40.734.1Mistral Large 4
Human preference59.7–not comparable
Multimodal58.0–not comparable
Long context65.3–not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningMathsKnowledgeMistral Large 4, Reasoning: 46.2Mistral Large 4, Maths: 54.2Mistral Large 4, Knowledge: 40.7Mistral Large 4.0 OSS, Reasoning: 64.2Mistral Large 4.0 OSS, Maths: 66.4Mistral Large 4.0 OSS, Knowledge: 34.1

Not drawn, measured for fewer than two of these models: Coding, Agents & tools, Instruction following, Long context, Multimodal, Human preference.

Model, click to add or removeIndex

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureMistral Large 4Mistral Large 4.0 OSS (high)
Blended price per 1M tokens$1.05–
Input / output per 1M$0.700 / $2.10– / –
Cached input per 1M$0.070–
Price per task: chat reply$0.0009–
Price per task: summarize a 30-page report$0.0097–
Price per task: code edit$0.0074–
Price per task: agentic coding session$0.050–
Price per task: structured extraction$0.0018–
Agentic coding session, input mostly cached$0.022–
Cost per benchmark task (AA Index)$4.21–
Output tokens per task (AA Index)135k–
Of which reasoning102k–
Time per task (AA Index)23 min–
Output speed97 tok/s–
Time to first answer40.9 s–
Time to a complete answer, last measured 13 Nov 2026119.0 s–

Each model is compared at its best-ranked reasoning-effort setting. Mistral Large 4 scores 52.9 and takes 41 s to its first answer token; Mistral Large 4 (high) scores 48.3 in 45 s, Mistral Large 4 (medium) scores 45.9 in 27 s. Mistral Large 4.0 OSS (high). Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

111 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherMistral Large 4Mistral Large 4.0 OSS (high)
BenchLeader IndexBenchLeader52.9–
Agents & tools scoreBenchLeader48.8–
Coding scoreBenchLeader53.8–
Human preference scoreBenchLeader59.7–
Knowledge scoreBenchLeader40.734.1
Long context scoreBenchLeader65.3–
Maths scoreBenchLeader54.266.4
Multimodal scoreBenchLeader58.0–
Reasoning scoreBenchLeader46.264.2
Blended price $/MBenchLeader$1.05–
Output speedBenchLeader97 tok/s–
Time to first answerBenchLeader40.9 s–
Context windowBenchLeader1,048,576 tokens–
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub88.1%†90.2%
Arena Hard PromptsArena (LMArena)1453–
LiveBench 2.0 ReasoningLiveBench85.2%†–
LiveBench 1.0 ReasoningLiveBench83.9%†–
Humanity's Last Exam (AA)Artificial Analysis35.0%–
HLE-Diamond (AA)Artificial Analysis15.1%–
CritPtArtificial Analysis10.6%–
CritPt v2 hard subset (AA)Artificial Analysis24.5%–
MysteryMechanismVals AI12.2%†–
Arena ExpertArena (LMArena)1464–
Chess PuzzlesEpoch AI Benchmarking Hub33.0%†29.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub17.0%†20.0%
LMCAEpoch AI Benchmarking Hub35.8%–
DTBenchEpoch AI Benchmarking Hub88.2%–
Coding
SciCodeSciCode54.2%–
FrontierCodeCognition39.4%†–
Arena CodingArena (LMArena)1484–
Arena WebDevArena (LMArena)1565–
LiveBench 2.0 CodingLiveBench79.4%†–
LiveBench 1.0 CodingLiveBench77.2%†–
IOIVals AI45.3%†–
Code MigrationVals AI30.6%†–
SWE-bench MultimodalVals AI33.1%†–
Vibe Code Bench 1-100Vals AI14.3%†–
Vibe Code Bench v1.1Vals AI78.4%†–
Show all 111 benchmarks
MetricPublisherMistral Large 4Mistral Large 4.0 OSS (high)
Coding
SWE Atlas v1.1Scale AI SEAL45.2%–
SWE Atlas v1.1: Codebase QnAScale AI SEAL61.3%–
SWE Atlas v1.1: Test WritingScale AI SEAL40.0%–
SWE Atlas v1.1: RefactoringScale AI SEAL34.3%–
MirrorCodeEpoch AI Benchmarking Hub40.0%†–
Surface Evolver BenchEpoch AI Benchmarking Hub28.1%–
ALE-BenchEpoch AI Benchmarking Hub1213.2–
WeirdML v3WeirdML4.5%†–
Terminal-Bench 4.0 (AA)Artificial Analysis26.8%–
GameBench 2 by reasoning settingDuelLab GameBench 257.6†–
Design Arena WebsiteDesign Arena1258–
FrontierCode 1.1 (AA)Artificial Analysis41.5%–
FrontierCode 2.0 (AA)Artificial Analysis13.6%–
Agents & tools
Terminal-BenchTerminal-Bench27.6%†–
APEX-AgentsMercor51.6%†–
Arena AgentArena (LMArena)-5.4–
LiveBench Long AgenticLiveBench24.7%†–
LiveBench BrowserLiveBench17.4%†–
LiveBench Agentic CodingLiveBench57.2%†–
GDPval-AA v2.1, normalised scoreArtificial Analysis46.1%–
GDPval-AA v2.1Artificial Analysis1431–
Finance Agent v2Vals AI54.7%†–
Finance Agent v2.1Vals AI61.4%†–
Harvey's Legal Agent Benchmark v1.1Vals AI17.5%†–
Harvey's Legal Agent Benchmark v1Vals AI15.8%†–
Legal Research BenchVals AI31.7%†–
Tax Agent BenchVals AI30.4%†–
Terminal-Bench 4.0 (Vals)Vals AI27.8%†–
Arena Agent Task CompletionArena (LMArena)-4.7–
AutomationBenchArtificial Analysis59.9%–
GDP.pdf (AA)Artificial Analysis18.6%–
Harvey LAB-AA v1.1Harvey3.1%–
AA-Briefcase v1.1Artificial Analysis1393–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis4.8%–
CWE-Bench (AA)Artificial Analysis50.8%–
DeepSec-Bench (AA)Artificial Analysis16.0%–
CyberGym end-to-end (AA)Artificial Analysis81.7%–
MKA (AA)Artificial Analysis35.1%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub58.6%†–
FrontierMath Tier 4Epoch AI Benchmarking Hub26.8%†–
OTIS Mock AIMEEpoch AI Benchmarking Hub98.3%†99.4%
ProofBenchVals AI10.0%–
LiveBench MathematicsLiveBench93.6%†–
Arena MathsArena (LMArena)1459–
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub20.0%†22.4%
LiveBench 2.0 Data AnalysisLiveBench76.6%†–
LiveBench ResearchLiveBench80.3%†–
LiveBench 1.0 Data AnalysisLiveBench76.5%†–
AA-Omniscience v1.1Artificial Analysis-7.5–
BioMysteryBenchVals AI67.0%†–
Excel Modeling BenchmarkVals AI56.2%†–
Excel Modeling Benchmark v1.1Vals AI69.0%†–
MedCodeVals AI40.7%†–
MedScribeVals AI80.4%†–
Immigration BenchVals AI63.7%†–
EBR-benchEpoch AI Benchmarking Hub–18.1%
AA-Omniscience v1.1: accuracyArtificial Analysis23.9%–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis58.7%–
Instruction following
LiveBench LanguageLiveBench49.6%†–
LiveBench Instruction FollowingLiveBench64.8%†–
Arena Instruction FollowingArena (LMArena)1418–
Human preference
Arena TextArena (LMArena)1420–
Arena Creative WritingArena (LMArena)1359–
Arena Multi-turnArena (LMArena)1429–
Arena Longer QueriesArena (LMArena)1420–
Multimodal
Arena VisionArena (LMArena)1244–
LiveBench DesignLiveBench76.3%†–
MMMU-ProArtificial Analysis76.4%–
Furniture AssemblyEpoch AI Benchmarking Hub20.8%†30.8%
VIALS-AAHandshake AI20.5%†–
Code2Video-AAArtificial Analysis931–
AA-3DArtificial Analysis882–
Long context
AA-LCR v1.1Artificial Analysis81.3%–
MLCRArtificial Analysis7.8%–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub154.6–
LiveBench 2.0LiveBench60.3%†–
LiveBench 1.0LiveBench71.8%†–
AA Intelligence Index v5.0.1Artificial Analysis25.3–
AA Intelligence Index v4.3.2Artificial Analysis38.4–
AA Progress Index v1Artificial Analysis312.4–
Vals Finance IndexVals AI56.9%†–
Vals Legal IndexVals AI65.6%†–
AA Cyber IndexArtificial Analysis49.5–
Safety & honesty
FORTRESSScale AI SEAL23.1%–
DistressBenchScale AI SEAL77.0%–
Metadata
AA Openness IndexArtificial Analysis38.9†–

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • Mistral Large 4: BenchLeader Index 52.2 → 52.9.
  • Mistral Large 4: 1 new result (GDPval-AA v2.1).
  • Mistral Large 4: 1 score revision.
  • Mistral Large 4: 2 score revisions.
  • Mistral Large 4: 6 score revisions.
  • Mistral Large 4: BenchLeader Index 52.7 → 52.2.
  • Mistral Large 4: 7 score revisions.
  • Mistral Large 4: 8 score revisions.
  • Mistral Large 4: 2 score revisions.
  • Mistral Large 4: 1 score revision.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

Mistral Large 4 vs Mistral Large 4.0 OSS: questions

When was Mistral Large 4.0 OSS released?
Mistral Large 4.0 OSS was released on 29 Oct 2026.