Benchmark Almanac

WorkflowArena

Claude Opus 5.5 37% leads since November 2026
Not in model cards

Agents write executable workflows that coordinate other agents on professional tasks. Hidden verifiers check the completed deliverables.

The overall resolved rate averages the eight domain rates equally. Each task requires every acceptance criterion to pass.

Released
November 2026
Kind
Multi-agent games
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions, Text, 3D
Interaction
Multi-agent
Environment
Terminal
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-6 Astra #4 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
9
Entries in its paper, 5 models we track

Claude Opus 5.5 leads with 37% among models released by September 2026 Claude Opus 5.5 leads with 37% among models released by September 2026

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran on its own with several settings, we keep its best. Every run is in the next chart. Agents other companies built around a model get a pale shade of that model's lab color. Gray marks blends and entries not in our model list, such as fine-tunes and older models. None of these count toward leaders. Agents built around a model, such as Claude Code + Kimi K3.1 Pro, are their own entries. We don't count them as the model's score. This view shows only models we track, each at its best.

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

  1. 1 Claude Opus 5.5 max 37%
  2. 2 Claude Sonnet 5.5 max 33.2%
  3. 3 GPT-6 Astra xhigh 30.8%
  4. 4 GPT-6.1 Sol xhigh 29.5%
  5. 5 Claude Code + Kimi K3.1 Pro max 16.5%
  6. 6 GPT-6 Luna max 13.3%
  7. 7 Claude Code + Doubao Seed Evolving Nov 2026, high 10.8%
  8. 8 Claude Code + DeepSeek V4.1 Flash max 6.08%
  9. 9 Claude Code + GLM-5.3-Flash max 4.43%

Source: The WorkflowArena paper

#EntryScoreSettingReleasedRun onSource
1 Claude Opus 5.5 37% max, Claude Code Sep 2026 Nov 2026 Official leaderboard
2 Claude Sonnet 5.5 33.2% max, Claude Code Sep 2026 Nov 2026 Official leaderboard
3 GPT-6 Astra 30.8% xhigh, Codex Sep 2026 Nov 2026 Official leaderboard
4 GPT-6.1 Sol 29.5% xhigh, Codex Sep 2026 Nov 2026 Official leaderboard
5 Claude Code + Kimi K3.1 Pro 16.5% max Oct 2026 Nov 2026 Official leaderboard
6 GPT-6 Luna 13.3% max, Codex Sep 2026 Nov 2026 Official leaderboard
7 Claude Code + Doubao Seed Evolving Nov 2026 10.8% high Nov 2026 Nov 2026 Official leaderboard
8 Claude Code + DeepSeek V4.1 Flash 6.08% max Sep 2026 Nov 2026 Official leaderboard
9 Claude Code + GLM-5.3-Flash 4.43% max Aug 2026 Nov 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 Claude Opus 5.5 37% max, Claude Code Sep 2026 Nov 2026 Official leaderboard
2 Claude Sonnet 5.5 33.2% max, Claude Code Sep 2026 Nov 2026 Official leaderboard
3 GPT-6 Astra 30.8% xhigh, Codex Sep 2026 Nov 2026 Official leaderboard
4 GPT-6.1 Sol 29.5% xhigh, Codex Sep 2026 Nov 2026 Official leaderboard
5 GPT-6 Luna 13.3% max, Codex Sep 2026 Nov 2026 Official leaderboard

Rankings by part

We read 16 parts from the benchmark’s paper in 2 groups: resolved rate and continuous score. Each ranks models on one part of the test. Claude Opus 5.5 leads 12 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Software & systems

5 models we track have a score here.

  1. 1 Claude Opus 5.5 29.4%
  2. 2 Claude Sonnet 5.5 11.8%
  3. 3 GPT-6 Astra 5.88%
  4. 4 GPT-6 Luna ↑1 0%
  5. 5 GPT-6.1 Sol ↓1 0%

More effort adds a median 15 points

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Claude Opus 5.5 Claude Code
low medium high max
37%
Claude Sonnet 5.5 Claude Code
low medium high xhigh max
33.2%
GPT-6 Astra Codex
low high max
30.8%
GPT-6.1 Sol Codex
low high xhigh
29.5%

Source: The WorkflowArena paper

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 max Claude Code 37%
Claude Opus 5.5 high Claude Code 34.1% $6.66
Claude Opus 5.5 xhigh Claude Code 33.2%
Claude Sonnet 5.5 max Claude Code 33.2%
GPT-6 Astra xhigh Codex 30.8% $29
GPT-6 Astra high Codex 29.8%
GPT-6.1 Sol xhigh Codex 29.5%
Claude Opus 5.5 medium Claude Code 29.1%
GPT-6 Astra medium Codex 28.1%
GPT-6.1 Sol max Codex 28%
GPT-6 Astra low Codex 27.8%
GPT-6.1 Sol low Codex 27.6% $3.61
GPT-6 Astra max Codex 27.3%
GPT-6.1 Sol medium Codex 27.3%
GPT-6.1 Sol high Codex 25.6%
Claude Sonnet 5.5 high Claude Code 25.3%
Claude Sonnet 5.5 xhigh Claude Code 23.2%
Claude Opus 5.5 low Claude Code 22%
Claude Sonnet 5.5 medium Claude Code 19.8%
Claude Sonnet 5.5 low Claude Code 16.9%
Claude Code + Kimi K3.1 Pro max Claude Code 16.5%
GPT-6 Luna max Codex 13.3%
Claude Code + Doubao Seed Evolving high Claude Code 10.8%
Claude Code + DeepSeek V4.1 Flash max Claude Code 6.08%
Claude Code + GLM-5.3-Flash max Claude Code 4.43%

More effort adds a median 15 points

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Claude Opus 5.5 Claude Code
low medium high max
37%
Claude Sonnet 5.5 Claude Code
low medium high xhigh max
33.2%
GPT-6 Astra Codex
low high max
30.8%
GPT-6.1 Sol Codex
low high xhigh
29.5%

Source: The WorkflowArena paper

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Opus 5.5 max Claude Code 37%
Claude Opus 5.5 high Claude Code 34.1% $6.66
Claude Opus 5.5 xhigh Claude Code 33.2%
Claude Sonnet 5.5 max Claude Code 33.2%
GPT-6 Astra xhigh Codex 30.8% $29
GPT-6 Astra high Codex 29.8%
GPT-6.1 Sol xhigh Codex 29.5%
Claude Opus 5.5 medium Claude Code 29.1%
GPT-6 Astra medium Codex 28.1%
GPT-6.1 Sol max Codex 28%
GPT-6 Astra low Codex 27.8%
GPT-6.1 Sol low Codex 27.6% $3.61
GPT-6 Astra max Codex 27.3%
GPT-6.1 Sol medium Codex 27.3%
GPT-6.1 Sol high Codex 25.6%
Claude Sonnet 5.5 high Claude Code 25.3%
Claude Sonnet 5.5 xhigh Claude Code 23.2%
Claude Opus 5.5 low Claude Code 22%
Claude Sonnet 5.5 medium Claude Code 19.8%
Claude Sonnet 5.5 low Claude Code 16.9%
GPT-6 Luna max Codex 13.3%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Benchmark paper 1 leader

Source: The WorkflowArena paper

Scores over time

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

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) Claude Opus 5.5 GPT-6 Astra 0% 20% 40% 60% 80% 100% Near the ceiling (85%) Claude Opus 5.5

Source: Official WorkflowArena leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Games and long tasks

All 261 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%

Data audit

All 11 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the benchmark’s paper (9 entries).
  • Pass Models scored 5 models we track, plus 4 other entries on its leaderboard.
  • Pass Newest score The newest is from September 2026, for GPT-6.1 Sol.
  • Pass Frontier models 4 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Pass Headroom The best is 37%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on lab.einsia.ai: 25 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check