Benchmark Almanac

QwenClawBench

Claude Sonnet 4.6 60.5% leads since April 2026
Qwen3.7-Max 64.3% leads since May 2026
Fading

Agents complete 100 tasks across 8 domains in isolated OpenClaw workspaces drawn from real user patterns. Automated checks and an LLM judge score each task.

Qwen built it as an internal benchmark for Qwen3.6-Plus and released it afterward.

Released
April 2026
Kind
General assistants
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Strongest model tested
GPT-5.4 #27 on the capability index
Last in a launch post
June 2026, dropped before it saturated
Last model added
June 2026
Scores refreshed
November 12th, 2026
12
Models scored on its own leaderboard
6
Launch posts list it, from 1 lab

Claude Sonnet 4.6 leads with 60.5% among models released by April 2026

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

Qwen3.7-Max leads with 64.3%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

The newest model with a score here came out in April 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 60.5 Claude Sonnet 4.6
  2. 59.5 Claude Opus 4.6
  3. 58.7 GLM-5.1
  4. 57.4 Qwen 3.6 Plus
  5. 56.7 GPT-5.4
  6. 56.5 MiMo-V2-Pro
  7. 56.1 Claude Opus 4.5
  8. 54.9 GLM-5
  9. 53.0 Qwen3.5-Plus
  10. 51.9 Kimi K2.5
  11. 50.7 DeepSeek-V3.2
  12. 50.5 MiniMax-M2.7
  1. 1 Claude Sonnet 4.6 v1.1 60.5%
  2. 2 Claude Opus 4.6 v1.1 59.5%
  3. 3 GLM-5.1 v1.1 58.7%
  4. 4 Qwen 3.6 Plus v1.1 57.4%
  5. 5 GPT-5.4 v1.1, high 56.7%
  6. 6 MiMo-V2-Pro v1.1 56.5%
  7. 7 Claude Opus 4.5 v1.1, high 56.1%
  8. 8 GLM-5 v1.1 54.9%
  9. 9 Qwen3.5-Plus v1.1 53%
  10. 10 Kimi K2.5 v1.1 51.9%
  11. 11 DeepSeek-V3.2 v1.1 50.7%
  12. 12 MiniMax-M2.7 v1.1, high 50.5%

Source: Official QwenClawBench leaderboard

17 models. Scroll sideways for the rest.

  1. 64.3 Qwen3.7-Max
  2. 61.8 Qwen3.7-Plus
  3. 60.5 Claude Sonnet 4.6
  4. 59.5 Claude Opus 4.6
  5. 59.0 Qwen 3.6 Max (Preview)
  6. 58.7 GLM-5.1
  7. 57.4 Qwen 3.6 Plus
  8. 56.7 GPT-5.4
  9. 56.5 MiMo-V2-Pro
  10. 56.1 Claude Opus 4.5
  11. 54.9 GLM-5
  12. 53.4 Qwen3.6 27B
  13. 53.0 Qwen3.5-Plus
  14. 52.6 Qwen 3.6 35B-A3B
  15. 51.9 Kimi K2.5
  16. 50.7 DeepSeek-V3.2
  17. 50.5 MiniMax-M2.7
  1. 1 Qwen3.7-Max 64.3%
  2. 2 Qwen3.7-Plus 61.8%
  3. 3 Claude Sonnet 4.6 v1.1 60.5%
  4. 4 Claude Opus 4.6 v1.1 59.5%
  5. 5 Qwen 3.6 Max (Preview) 59%
  6. 6 GLM-5.1 v1.1 58.7%
  7. 7 Qwen 3.6 Plus v1.1 57.4%
  8. 8 GPT-5.4 v1.1, high 56.7%
  9. 9 MiMo-V2-Pro v1.1 56.5%
  10. 10 Claude Opus 4.5 v1.1, high 56.1%
  11. 11 GLM-5 v1.1 54.9%
  12. 12 Qwen3.6 27B 53.4%
  13. 13 Qwen3.5-Plus v1.1 53%
  14. 14 Qwen 3.6 35B-A3B 52.6%
  15. 15 Kimi K2.5 v1.1 51.9%
  16. 16 DeepSeek-V3.2 v1.1 50.7%
  17. 17 MiniMax-M2.7 v1.1, high 50.5%

Sources: Labs’ model cards and official QwenClawBench leaderboard

#ModelScoreSettingDateSource
1 Claude Sonnet 4.6 60.5% v1.1 Feb 2026 Official leaderboard
2 Claude Opus 4.6 59.5% v1.1 Feb 2026 Official leaderboard
3 GLM-5.1 58.7% v1.1 Apr 2026 Official leaderboard
4 Qwen 3.6 Plus 57.4% v1.1 Mar 2026 Official leaderboard
5 GPT-5.4 56.7% v1.1, high Mar 2026 Official leaderboard
6 MiMo-V2-Pro 56.5% v1.1 Mar 2026 Official leaderboard
7 Claude Opus 4.5 56.1% v1.1, high Nov 2025 Official leaderboard
8 GLM-5 54.9% v1.1 Feb 2026 Official leaderboard
9 Qwen3.5-Plus 53% v1.1 Feb 2026 Official leaderboard
10 Kimi K2.5 51.9% v1.1 Jan 2026 Official leaderboard
11 DeepSeek-V3.2 50.7% v1.1 Dec 2025 Official leaderboard
12 MiniMax-M2.7 50.5% v1.1, high Mar 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Qwen3.7-Max 64.3% May 2026 Lab model card
2 Qwen3.7-Plus 61.8% Jun 2026 Lab model card
3 Claude Sonnet 4.6 60.5% v1.1 Feb 2026 Official leaderboard
4 Claude Opus 4.6 59.5% v1.1 Feb 2026 Official leaderboard
5 Qwen 3.6 Max (Preview) 59% Apr 2026 Lab model card
6 GLM-5.1 58.7% v1.1 Apr 2026 Official leaderboard
7 Qwen 3.6 Plus 57.4% v1.1 Mar 2026 Official leaderboard
8 GPT-5.4 56.7% v1.1, high Mar 2026 Official leaderboard
9 MiMo-V2-Pro 56.5% v1.1 Mar 2026 Official leaderboard
10 Claude Opus 4.5 56.1% v1.1, high Nov 2025 Official leaderboard
11 GLM-5 54.9% v1.1 Feb 2026 Official leaderboard
12 Qwen3.6 27B 53.4% Apr 2026 Lab model card
13 Qwen3.5-Plus 53% v1.1 Feb 2026 Official leaderboard
14 Qwen 3.6 35B-A3B 52.6% Apr 2026 Lab model card
15 Kimi K2.5 51.9% v1.1 Jan 2026 Official leaderboard
16 DeepSeek-V3.2 50.7% v1.1 Dec 2025 Official leaderboard
17 MiniMax-M2.7 50.5% v1.1, high Mar 2026 Official leaderboard

Labs report scores for 6 models in their cards. Switch the scores above to rank them together.

Number one over time 2 models have held number one with claims included since its release in April 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

Anthropic Qwen
Own leaderboard Not tested since Apr 2026
Claude Sonnet 4.6 60.5%
Lab claims Not tested since Jun 2026
Qwen3.7-Max 64.3%

Sources: Official QwenClawBench leaderboard and labs’ model cards

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Qwen3.7-Max Claude Sonnet 4.6 Claude Opus 4.5 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Qwen3.7-Max Claude Opus 4.5

Sources: Official QwenClawBench leaderboard and labs’ model cards

Labs headline it less often now, last in June 2026

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Qwen
DateLaunch postScoreAs the post writes it
Jun 1, 2026 Qwen3.7-Plus 61.8 Qwenclaw
May 20, 2026 Qwen3.7-Max 64.3 Qwenclaw
Apr 22, 2026 Qwen3.6 27B 53.4 QwenClawBench
Apr 20, 2026 Qwen 3.6 Max (Preview) 59 QwenClawBench Real-World Agent
Apr 15, 2026 Qwen 3.6 35B-A3B 52.6 QwenClawBench
Apr 2, 2026 Qwen 3.6 Plus 57.2 QwenClawBench

One lab claim next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. Every benchmark.

Model Lab says Board says Gap
Qwen 3.6 Plus 57.2% QwenClawBench 57.4% Leaderboard, v1.1 −0.2 pts

More in Tool use

All 144 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Carbon 53.9%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%