BlogLLM Benchmark Scores as Published, Side by Side
Mistral Large 4 (preview): Benchmark Scores as Published
Every result the catalog holds for Mistral Large 4 (preview): 24 tests in 8 areas, from 4 pages. 20 printed by Mistral AI itself, the rest by the company named beside each number. Each number is as printed; nothing is ranked or rescored.
Developer Mistral AIReleased 2026-10-06Data version 2026103001, as of 2026-10-30
Knowledge Work
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| AutomationBench | 59.9% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| GDPval-AA v2 | 1424 | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | - |
| Vals Finance Agent v2 | 54.7% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Harvey's Legal Agent Benchmark | 15.8% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| AA-Briefcase | 1393 | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| FinWorkBench (Finch) | 67.4% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
Agentic Coding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| DeepSWE v1.1 | 61.7% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15; printed on 1 more page | Harness: Mistral Vibe |
| Terminal-Bench 4.0 | 28.3% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| SWE-Atlas-QnA (setting not stated) | 59.4% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15; printed on 1 more page | - |
| Artificial Analysis Coding Agent Index (version not stated) | 49.8 | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
Science and Math
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| SciCode | 54.2% | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | - |
| SciCode-Verified | 91.8% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | Attempt budget: pass@1 (n=6) |
Multimodal Understanding
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| GDP.pdf | 18.6% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| ChartQA Pro | 63.1% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Dense200 (bbox) | 42% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
Cybersecurity
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Artificial Analysis Cyber Index (successes) | 50 | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| CyberGym-E2E | 82% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Cybench | 93% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
General Knowledge and Reasoning
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| AA-Omniscience Index | -5 | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | - |
Safety and Alignment
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| B3 AI Security Benchmark (attack resistance) | 93.3% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Cyber refusal rate (average of JailbreakBench, StrongREJECT, AgentHarm) | 95.3% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| KORA Benchmark (0 to 2) | 1.691 | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
Preference and Composite Rankings
| Test | Value | Printed by | Setup as printed |
|---|---|---|---|
| Artificial Analysis Intelligence Index | 38 | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | - |
| Surge AI human eval on code (1 to 5) | 3.74 | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
Price
API list prices in US dollars per one million tokens, as printed on the developer's pricing page on the day checked. Cost per task is measured by Vals AI.
| Price | Value | Printed by | Setup as printed |
|---|---|---|---|
| Input | $0.68 | Mistral AI, Mistral API pricingchecked 2026-10-15 | Offer: promotional price |
| Cached input | $0.07 | Mistral AI, Mistral API pricingchecked 2026-10-15 | Offer: promotional price |
| Output | $2.09 | Mistral AI, Mistral API pricingchecked 2026-10-15 | Offer: promotional price |
| Cost per taskVals AI, USD, lower is better | $39.46 | Vals AI, Vibe Code Bench v1.1 leaderboardchecked 2026-10-27 | Harness: OpenHands |
More from Mistral AI
Every number here is read from the same catalog file as the benchmark table, data version 2026103001. Where one page printed the model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.