gpt-oss-20b

Composite indexes

• Artificial Analysis Intelligence Index#105 of 142
  9independentAA Intelligence Index v4.3.2, highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Opus 5.5 at 58

Coding

• Terminal-Bench 4.0#61 of 88
  0%independentTerminal-Bench 4.0 (AA harness), highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Mythos 5.1 at 60.9%

• Terminal-Bench 2.x#87 of 103
  13.9%independentTerminal-Bench 2.1 (AA harness), highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Fable 5.1 at 91.4%

• SciCode#72 of 94
  38.9%independenthighArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Opus 5.5 at 66.9%

• LiveCodeBench#11 of 56
  77.7%independentAA, highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Fugu at 92.9%

Agents and tool use

• GDPval-AA#85 of 103
  327independentGDPval-AA, highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Opus 5.5 at 1846

• τ²-Bench#40 of 88
  60.2%independenttelecom (AA), highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: GLM-5.2 at 99.1%

Reasoning and knowledge

• Humanity’s Last Exam#95 of 147
  11%independentno tools (AA), highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Claude Opus 5.5 at 61.4%

• GPQA Diamond#100 of 142
  68.8%independentAA, highArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: GPT-6 Astra at 96.1%

• CritPt#66 of 139
  1.4%independenthighArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: GPT-5.6 Sol at 32.3%

Vision and long context

• AA Long Context Reasoning#102 of 139
  34.7%independenthighArtificial Analysis: gpt-oss-20b (high) ↗

  Leader: Kimi K3 at 88.7%

Human preference

• LMArena Text#54 of 54
  1317independentgpt-oss-20b, style control default, vote cutoff 2026-09-13LMArena Text leaderboard ↗

  Leader: Claude Fable 5 at 1506

Other published results