Mistral Medium 3.5

Composite indexes

• Artificial Analysis Intelligence Index#79 of 142
  14independentIntelligence Index v4.3; high effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#69 of 94
  40.2%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 4.0#75 of 88
  0%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Mythos 5.1 at 60.9%

• Terminal-Bench 2.x#61 of 103
  50.6%independentTerminal-Bench 2.1; high effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Fable 5.1 at 91.4%

• SWE-bench Verified#5 of 16
  77.6%vendorreasoning_effort highMistral Medium 3.5 model card ↗

  Leader: Claude Sonnet 5 at 85.2%

Agents and tool use

• τ²-Bench#14 of 88
  94.2%independenttelecom; high effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: GLM-5.2 at 99.1%

• GDPval-AA#66 of 103
  747independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Opus 5.5 at 1846

Reasoning and knowledge

• GPQA Diamond#90 of 142
  74.8%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#82 of 147
  13.8%independentno tools; high effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#104 of 139
  0%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: GPT-5.6 Sol at 32.3%

Vision and long context

• AA Long Context Reasoning#76 of 139
  69.3%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Kimi K3 at 88.7%

• MMMU-Pro#49 of 72
  64.9%independenthigh effortArtificial Analysis: Mistral Medium 3.5 ↗

  Leader: Claude Opus 5.5 at 87.7%

Other published results