Mistral AIMistral Large 3
Mistral Large 3 is a Mistral AI open-weights model, released 2 Dec 2025. Its best configuration ranks #512 of 785 on the BenchLeader Index at 45.4 ±3.6, in the lower half. The ± is the point: 292 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (59) and lowest in coding (38). At $0.750 per million tokens blended it is mid-priced. Output speed of 79 tokens per second puts it slower than most, with a first answer in 1.1 s. Last measured 22 Oct 2026.
- Blended price
- $0.750/M
- $0.500 in · $1.50 out
- Output speed
- 79 tok/s
- measured by Artificial Analysis
- First answer
- 1.14 s
- first token 0.62 s
- Context
- 262k
- Full answer
- 7.43 s
- median, reasoning included
- Cost per run
- $0.031
- one full Intelligence Index run
- Released
- 2 Dec 2025
- Overall index45
- Reasoning47
- Coding38
- Agents & tools38
- Maths50
- Knowledge49
- Instruction following39
- Human preference59
- Multimodal46
- Long context42
- Composite40
Versions
Mistral AI has shipped 5 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Mistral Large 4 | 6 Oct 2026 | 56.6 | #196 |
| Mistral Large 3this page | 2 Dec 2025 | 45.4 | #512 |
| Mistral Large 2.1 | 18 Nov 2024 | 42.5 | #590 |
| Mistral Large 2 | 24 Jul 2024 | 40.7 | #652 |
| Mistral Large 24.02 | 26 Feb 2024 | 40.2 | #660 |
How its index has moved
8 Sept 2026 to 21 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Hard Prompts | 1433#132 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 68.0%#317 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 4.2%#531 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | 68.4%#94 | Vals AI | |
| Kagi LLM Benchmark | 50.9%#83 | Kagi LLM Benchmark | |
| CritPt | 0.0%#315 | Artificial Analysis | |
| LMCAnot in index | 16.7%#168 | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | 65.1%#168 | Epoch AI Benchmarking Hub |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| SciCode | 36.6%#181 | SciCode | |
| Arena Coding | 1471#116 | Arena (LMArena) | |
| Arena WebDev | 1232#130 | Arena (LMArena) | |
| SciCode (AA)not in index | 36.6%#193 | Artificial Analysis | |
| LiveCodeBench | 55.3%#107 | Vals AI | |
| SWE-bench (Vals)not in index | 41.4%#82 | Vals AI | |
| ALE-Benchnot in index | 264.7#120 | Epoch AI Benchmarking Hub | |
| Terminal-Bench 4.0 (AA)not in index | 0.0%#152 | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 12.0%#190 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Terminal-Bench Hard | 15.9%#190 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 24.6%#300 | Artificial Analysis | |
| τ³-Banking (AA) | 5.8%#173 | Artificial Analysis | |
| Terminal-Bench 2.0 (Vals)not in index | 9.0%#62 | Vals AI | |
| AutomationBenchnot in index | 1.5%#188 | Artificial Analysis | |
| GDP.pdf (AA)not in index | 1.2%#198 | Artificial Analysis |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AIME (Vals) | 42.9%#63 | Vals AI | |
| MGSM | 85.4%#64 | Vals AI | |
| Arena Maths | 1400#149 | Arena (LMArena) |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -39.6#320 | Artificial Analysis | |
| MMLU-Pro | 79.8%#88 | Vals AI | |
| LegalBench | 79.1%#95 | Vals AI | |
| CorpFin | 61.0%#60 | Vals AI | |
| TaxEval | 73.1%#52 | Vals AI | |
| MedQA | 82.2%#64 | Vals AI | |
| CaseLaw v2not in index | 61.4%#21 | Vals AI | |
| AA-Omniscience: accuracynot in index | 24.9%#234 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 14.0%#355 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| IFBench | 36.2%#304 | Artificial Analysis | |
| Arena Instruction Followingnot in index | 1402#135 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1414#134 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1375#141 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1420#128 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1418#135 | Arena (LMArena) |
Multimodal
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Vision | 1220#81 | Arena (LMArena) | |
| MMMU-Pro | 55.7%#225 | Artificial Analysis | |
| MMMU-Pro (Vals)not in index | 66.2%#72 | Vals AI | |
| MortgageTaxnot in index | 52.1%#78 | Vals AI | |
| SAGEnot in index | 24.6%#86 | Vals AI |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-LCR | 36.0%#374 | Artificial Analysis | |
| MLCRnot in index | 1.1%#165 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA Intelligence Index v4.3.2 | 9.3#404 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Mistral | 75 tok/s | 0.60 s | $0.550 | $1.65 | 262k | – |
| Mistral | 47 tok/s | 0.64 s | $0.500 | $1.50 | 262k | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.050 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0007 | $0.0005 | 4.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0069 | $0.0029 | 8.7 s |
| Code edit | 6,000 / 1,500 | $0.0053 | $0.0032 | 20.0 s |
| Agentic coding session | 60,000 / 4,000 | $0.036 | $0.016 | 51.5 s |
| Structured extraction | 2,000 / 200 | $0.0013 | $0.0006 | 3.7 s |
More from Mistral AI
See also
Questions people ask
- What are Mistral Large 3's benchmark scores?
- As of 24 Oct 2026, Mistral Large 3 scores 45.4 ±3.6 on the BenchLeader Index, strongest in human preference (59) and weakest in coding (38). That rests on 24 independent benchmark results, and every score on this page links to its publisher.
- How does Mistral Large 3 rank?
- Mistral Large 3 is #512 of 785 ranked configurations on 24 Oct 2026. 292 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Mistral Large 3 cost per 1M tokens?
- As of 24 Oct 2026, Mistral Large 3 costs $0.500 per 1M input tokens and $1.50 per 1M output tokens, $0.750 blended (three input tokens per output token).
- When was Mistral Large 3 released?
- Mistral AI released Mistral Large 3 on 2 Dec 2025.
Data as of 24 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Mistral Large 3: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/mistral-large-3.