Mistral Large 4

Leaderboards

Where Mistral Large 4 sits among the top 20 in each category.

By task

Scored from votes on each kind of prompt alone, on the same 1,000-point scale.

Test results

Every test Mistral Large 4 has a published result on, out of 100%, with its rank among the models that took it.

• DeepSWE 1.1

  #32 / 50

  DeepSWE 1.1 evaluates software engineering agents using the mini-swe-agent harness.

  61.7%

  Self-reported · In the coding score

• SciCode

  #1 / 30

  SciCode is a research coding benchmark curated by scientists that challenges language models to code solutions for scientific problems. It contains 338 subproblems decomposed from 80 challenging main problems across 16 natural science sub-fields including mathematics, physics, chemistry, biology, and materials science. Problems require knowledge recall, reasoning, and code synthesis skills.

  91.8%

  Self-reported

• Finance Agent v2

  #5 / 28

  Finance Agent v2 is an agentic financial-analysis benchmark from Vals that evaluates models on real-world finance workflows, measuring their ability to retrieve and reason over financial documents, perform multi-step calculations, and produce accurate analyses.

  54.7%

  Self-reported

• Terminal-Bench 4.0

  #20 / 28

  Terminal-Bench 4.0 evaluates AI agents on real-world work in terminal and command-line containerized environments across 66 tasks, with emphasis on science-adjacent and frontier engineering problems. Relative to earlier Terminal-Bench releases, 4.0 increases timeouts and adaptively raises RAM/CPU on selected tasks to reduce harness and resource confounds.

  28.3%

  Self-reported · In the coding score· In the agents score

• AutomationBench

  #2 / 24

  AutomationBench is a tool-use benchmark that evaluates AI agents on automating real-world workflows, testing their ability to orchestrate tools and complete multi-step automation tasks.

  59.9%

  Self-reported · In the agents score

• CyberGym

  #10 / 21

  CyberGym is a benchmark for evaluating AI agents on cybersecurity tasks, testing their ability to identify vulnerabilities, perform security analysis, and complete security-related challenges in a controlled environment.

  82.0%

  Self-reported

• GDP.pdf

  #14 / 14

  GDP.pdf is a knowledge-work vision benchmark that evaluates models on economically valuable professional tasks presented as visual documents (PDFs), testing document-based reasoning, chart and table interpretation, and problem solving without tools.

  18.6%

  Self-reported

• SWE Atlas - Codebase QnA

  #1 / 5

  SWE Atlas - Codebase QnA evaluates a model's ability to answer questions about real codebases, measuring repository-level comprehension and the ability to reason about code structure, behavior, and intent across an entire project.

  59.4%

  Self-reported

• AA-Briefcase

  #3 / 4

  AA-Briefcase is an Artificial Analysis evaluation of AI systems on professional knowledge-work tasks, reported as an Elo score.

  46.4%

  Self-reported

• ChartQAPro

  #4 / 4

  ChartQAPro is a challenging benchmark for question answering over diverse, real-world charts and infographics.

  63.1%

  Self-reported

• Harvey's Legal Agent Benchmark

  #3 / 4

  Harvey's Legal Agent Benchmark evaluates AI agents on complex legal workflows and reports the all-pass rate.

  15.8%

  Self-reported

• CyBench

  #2 / 3

  CyBench is a suite of Capture-the-Flag (CTF) challenges measuring agentic cyber attack capabilities. It evaluates dual-use cybersecurity knowledge and measures the 'unguided success rate', where agents complete tasks end-to-end without guidance on appropriate subtasks.

  93.0%

  Self-reported

• B3 AI Security Benchmark

  #1 / 1

  B3 AI Security Benchmark evaluates agent-backbone robustness against contextual prompt-injection attacks collected in Lakera's Agent Breaker Challenge.

  93.3%

  Unverified

• Coding Agent Index

  #1 / 1

  Coding Agent Index is a combined coding-agent score (as reported by Mistral / Artificial Analysis) summarizing performance across software-engineering agent evaluations such as DeepSWE, SWE-Atlas QnA, and Terminal-Bench.

  49.8%

  Unverified

• Dense200

  #1 / 1

  Dense200 is a visual grounding / bounding-box benchmark that evaluates how accurately models localize objects in dense natural scenes.

  42.0%

  Unverified

• Finch (FinWorkBench)

  #1 / 1

  Finch (FinWorkBench) evaluates model capabilities at creating and editing spreadsheets on real-life finance and accounting use cases.

  67.4%

  Unverified

• KORABench

  #1 / 1

  KORABench is a nonprofit, independent, open-source benchmark for child-safety behavior in AI systems, measuring model behavior against 26 child-specific risks.

  84.5%

  Unverified

• Surge AI Coding Quality

  #1 / 1

  Surge AI Coding Quality is a blinded human evaluation where professional annotators rate model coding outputs on a 1-5 scale with model identities hidden.

  74.8%

  Unverified