Gemini 3.8 Flash

Composite indexes

• Artificial Analysis Intelligence Index#26 of 151
  41independentv4.3.2, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗
  40independentv4.3.2, medium thinkingArtificial Analysis: Gemini 3.8 Flash (medium) ↗
  34independentv4.3.2, low thinkingArtificial Analysis: Gemini 3.8 Flash (low) ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#19 of 100
  56.6%independentAA run, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 4.0#29 of 96
  19.7%independentAA run (Terminal-Bench 4.0), high thinkingArtificial Analysis: Gemini 3.8 Flash ↗
  19.1%independenthigh thinking, mini-SWE-agent harness, ±3.4Terminal-Bench 4.0 official leaderboard ↗

  Leader: Claude Sonnet 5.5 at 70.6%

• Terminal-Bench 2.x#10 of 105
  89.4%vendorTerminal-Bench 2.1, Terminus 2 harness, self-computedGemini 3.8 Flash model card (DeepMind) ↗
  87.6%independentAA run (Terminal-Bench 2.1), high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Claude Fable 5.1 at 91.4%

Agents and tool use

• GDPval-AA#33 of 109
  1545vendorGDPval-AA v2 (sourced from AA leaderboard)Gemini 3.8 Flash model card (DeepMind) ↗
  1412independentGDPval-AA v2.1, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Muse Wonder Full at 1857

• OSWorld-Verified#20 of 21
  59%vendorOSWorld 2.0 partial score, batch tool enabledGemini 3.8 Flash model card (DeepMind) ↗

  Leader: Kimi K3 at 84.8%

Reasoning and knowledge

• GPQA Diamond#5 of 144
  95.3%independentAA run, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#15 of 153
  47.8%independentAA run, no tools, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Claude Sonnet 5.5 at 64.5%

• CritPt#22 of 144
  18.3%independentAA run, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Muse Wonder Full at 33%

• ARC-AGI-2#7 of 25
  89.2%independenthigh thinking; ARC Prize verified, cost $0.40/taskARC Prize leaderboard data (semi-private) ↗

  Leader: GPT-6 Astra at 95%

• ARC-AGI-3#3 of 9
  10.4%independenthigh thinking; ARC Prize semi-private (35.0% with provider adapter)ARC Prize leaderboard data (semi-private) ↗

  Leader: GPT-6 Astra at 62.7%

Vision and long context

• AA Long Context Reasoning#24 of 144
  81.3%independentAA-LCR v1.1, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Kimi K3 at 88.7%

• MMMU-Pro#3 of 72
  85.6%independentAA run, high thinkingArtificial Analysis: Gemini 3.8 Flash ↗

  Leader: Claude Opus 5.5 at 87.7%

Human preference

• LMArena Text#7 of 57
  1493independentgemini-3.8-flash-high; 5,076 votesLMArena text leaderboard ↗

  Leader: Gemini 4 Argon at 1525

• LMArena WebDev#26 of 58
  1584independentgemini-3.8-flash-high; 6,645 votesLMArena Code Arena | WebDev leaderboard ↗

  Leader: Claude Opus 5.5 at 1818

Other published results