Mistral Large 4

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-10
RankedEnough results for an index.
IndexAll categories, out of 100.60.1±4.0Median 50.7
CoverageIndex weight that has results.85%
SpeedOutput tokens per second.57/sMedian 76/s
First tokenSeconds to the first output token.84.0 sMedian 1.1 s
Input / 1MUS dollars per 1M input tokens.$0.70Median $0.60
Output / 1MUS dollars per 1M output tokens.$2.10Median $2.50
ContextMaximum tokens in one request.1MMedian 262K
EloLMArena rating and rank.1413±7 #142Median 1425

Run locally

1000B parameters
16-bit
2101.0 GB
8-bit
1116.6 GB
4-bit
637.6 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory for a short context. 4-bit needs more than 128 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.16 counted65.544 / 221
CodingCode writing and repair.8 counted62.159 / 214
ReasoningLogic problems and puzzles.7 counted58.374 / 225
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted44.1169 / 266
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted57.2—
MathMath problems.5 counted60.747 / 171

Results

41 counted
BenchmarkResultThe score from the source.PlacePlace among results on this benchmark.
Artificial Analysis CyberGym-E2E-AAAgenticAbove fitFar above this model's index.79.4%2 / 21
OTIS Mock AIME 2024-2025Math98.9% ±0.425 / 203
Artificial Analysis Cyber IndexAgenticAbove fitFar above this model's index.50.9%5 / 21
GPQA diamond · Epoch AIKnowledge89.1% ±1.854 / 220
LiveBench MathematicsMath93.6%18 / 66
GDPval-AA normalizedAgentic46.1%37 / 132
BrowseCompAgentic84.5%18 / 63
Vibe Code Bench v1.1Coding78.4% ±3.540 / 130
Critical Physics TasksReasoning10.0%64 / 207
LiveBench Agentic CodingAgentic57.2%22 / 66
LiveBench Data AnalysisReasoning76.5%33 / 80
Terminal-Bench 4.0Agentic27.8% ±1.831 / 68
Terminal-Bench 4.1.0Agentic27.6% ±2.026 / 52
FrontierMath-Tiers-1-3-v2-PrivateMath58.6% ±2.948 / 93
Code MigrationCoding30.6% ±4.241 / 77
LiveBench ReasoningReasoning83.9%43 / 80
FrontierMath-Tier-4-v2-PrivateMath26.8% ±7.042 / 75
LiveBench CodingCoding77.2%37 / 66
Chess PuzzlesReasoning31.0% ±4.753 / 93
Mystery Game PuzzlesReasoning18.5% ±4.051 / 83
LiveBench Instruction FollowingInstruction64.8%42 / 66
MirrorCodeCoding40.0% ±10.626 / 40
IOICoding45.3% ±3.831 / 46
EBR-benchReasoning18.1% ±3.223 / 33
Agent Arena task outcomeAgentic−7.546 / 59
Agent Arena steerabilityAgentic−9.046 / 59
Vibe Code Bench 1-100Coding14.3% ±3.221 / 26
LiveBench DesignAgentic65.6%43 / 52
FrontierCode 2.0Coding19.7%27 / 32
CWE-bench v2Agentic45.6%17 / 20
SimpleQA VerifiedKnowledge21.2% ±1.380 / 94
Artificial Analysis GDP.pdfAgentic17.6%20 / 23
Furniture AssemblyReasoning25.8% ±6.035 / 40
LiveBench Coding 2.0Coding60.0%47 / 52
ProofBench v1.1Math10.0% ±3.050 / 55
LiveBench Browser UseAgentic8.0%45 / 49
LiveBench Long AgenticAgentic11.8%48 / 51
LiveBench LanguageKnowledge49.6%65 / 66
LiveBench ResearchAgentic28.1%52 / 52
HLAB-Vals-1.1Legal17.5% ±3.210 / 42
Finance Agent (v2)Finance54.7% ±0.626 / 79
Tax Agent BenchFinance30.4% ±3.336 / 89
MedScribeHealthcare80.4% ±2.059 / 111
MedCodeHealthcare40.7% ±2.262 / 108
Legal Research BenchLegal31.7% ±3.258 / 98
EMBFinance56.2% ±3.247 / 75
BioMysteryBenchScience67.0% ±2.723 / 34
MysteryMechanismScience12.2% ±2.234 / 34
CybenchAgentic93.0%—

Sources

BenchLMCC BY-NC 4.0Epoch AICC BYLiveBenchNo licence statedVals AINo licence statedTerminal-BenchNo licence statedLMArenaCC BY 4.0

History

13 releases
Index60.2 → 60.1
23 Oct 202621 Nov 2026
Output / 1M$2.09 → $2.10
23 Oct 202621 Nov 2026

Questions

How much does Mistral Large 4 cost?
$0.70 per 1M input tokens, $2.10 per 1M output tokens.
How good is Mistral Large 4?
Its index is 60.1 out of 100. That is place 73 of 282 ranked models.
What is Mistral Large 4 best at?
Agentic: 65.5 out of 100, place 44 of 221.
How fast is Mistral Large 4?
57 output tokens per second, 84.0 s to the first token.
What is the context window of Mistral Large 4?
1M tokens.
Is Mistral Large 4 open source?
Yes. Its weights are published.
When was Mistral Large 4 released?
October 2026.

Badge

Mistral Large 4 on Inference360: place 73 of 282, index 60.1

Compare with

GLM-5.259.9DeepSeek V4 Pro60.3DeepSeek V4 Flash Vision Exp59.6DeepSeek V4 Flash 073160.6Mistral Medium 3.542.4

Lower price

dots3-note Preview62.3 · FreeGLM-5.3-Flash62.4 · $0.078/1MHy4 flash64.5 · $0.14/1M

More from Mistral

All models
Pixtral Large 2502UnrankedLeanstral 1.5UnrankedMistral Medium 3.542.4Mistral Medium 3.5 128B49.7Mistral Small 438.0