Shared benchmarks

19
Shared benchmarks for Qwen3.6-Plus and Qwen3.7-Plus
BenchmarkQwen3.6-PlusQwen3.7-Plus
  • 4.3%
  • 3.7%
Evaluation details for AutomationBench 1.0.6

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on AutomationBench 1.0.6

Qwen3.7-Plus

Record for Qwen3.7-Plus on AutomationBench 1.0.6
  • 40.8%
  • 38.2%
Evaluation details for Finance Agent Benchmark 2 — October 2026 evaluation

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 88.4%
  • 81.8%none
  • 87.9%
Evaluation details for GPQA Diamond — Epoch AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness

Qwen3.7-Plus

Record for Qwen3.7-Plus on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for Qwen3.7-Plus on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
GPQAPotentially non-equivalent: Benchmark versions differ. Evaluator sets differ.
  • 87.4%Diamond — Vals AI harness · Vals AI
  • 90.3%Diamond · Qwen Team
Potentially non-equivalent · Evaluation details for GPQA

Benchmark versions differ. Evaluator sets differ. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness

Qwen3.7-Plus

Record for Qwen3.7-Plus on Graduate-Level Google-Proof Q&A Diamond
  • 1.3%
  • 0.0%
Evaluation details for HLAB October 2026 evaluation

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on Harvey's Legal Agent Benchmark October 2026 evaluation

Qwen3.7-Plus

Record for Qwen3.7-Plus on Harvey's Legal Agent Benchmark October 2026 evaluation
  • 74.2%
  • 79.1%
Evaluation details for IFBench Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on IFBench Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on IFBench Original
  • 88.0%
  • 90.3%
Evaluation details for MATH-Vision Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on MATH-Vision Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on MATH-Vision Original
  • 68.7%
  • 71.0%
Evaluation details for MedXpertQA MM

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on MedXpertQA MM

Qwen3.7-Plus

Record for Qwen3.7-Plus on MedXpertQA MM
  • 88.5%
  • 88.5%
Evaluation details for MMLU-Pro Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on MMLU-Pro Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on MMLU-Pro Original
  • 89.5%
  • 89.0%
Evaluation details for MMMLU Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on MMMLU Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on MMMLU Original
  • 62.5%
  • 73.3%
Evaluation details for OSWorld Verified

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on OSWorld Verified

Qwen3.7-Plus

Record for Qwen3.7-Plus on OSWorld Verified
  • 85.4%
  • 86.9%
Evaluation details for RealWorldQA Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on RealWorldQA Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on RealWorldQA Original
  • 41.4%
  • 51.3%
Evaluation details for SciCode Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on SciCode Original

Qwen3.7-Plus

Record for Qwen3.7-Plus on SciCode Original
  • 45.7%
  • 54.9%
Evaluation details for SkillsBench 78-task self-contained subset — OpenCode

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 73.8%
  • 75.8%
Evaluation details for SWE-bench Multilingual

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on SWE-bench Multilingual

Qwen3.7-Plus

Record for Qwen3.7-Plus on SWE-bench Multilingual
  • 78.8%
  • 77.7%
Evaluation details for SWE-bench Verified

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on SWE-bench Verified

Qwen3.7-Plus

Record for Qwen3.7-Plus on SWE-bench Verified
SWE-bench ProPotentially non-equivalent: Benchmark versions differ.
  • 56.6%Public — Qwen3.6 refined tasks · Qwen Team
  • 57.6%Public — Qwen refined tasks · Qwen Team
Potentially non-equivalent · Evaluation details for SWE-bench Pro

Benchmark versions differ. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on SWE-bench Pro Public — Qwen3.6 refined tasks

Qwen3.7-Plus

Record for Qwen3.7-Plus on SWE-bench Pro Public — Qwen refined tasks
  • 53.2%
  • 52.8%
Evaluation details for Terminal-Bench 2.1 — Vals AI Terminus 2

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Qwen3.6-Plus

Record for Qwen3.6-Plus on Terminal-Bench 2.1 — Vals AI Terminus 2

Qwen3.7-Plus

Record for Qwen3.7-Plus on Terminal-Bench 2.1 — Vals AI Terminus 2
  • 25.6%
  • 46.4%
Evaluation details for Vibe Code Bench 1.1 — OpenHands, October 2026 evaluation

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.