WildAWM validation
Agents complete 33 held-out computer-use tasks in generated software environments. Harbor packages the validation split separately from the training tasks.
The October 2026 paper reports 21.59% for supervised WildAWM and 0.38% for base Qwen3.5-4B in its four-billion-parameter validation comparison.
- Made by
- MSR Orchard
- Released
- October 2026
- Kind
- Desktop apps
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Desktop apps, APIs and tools, Terminal
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- Based on
- A version of another benchmark
- In views
- Computer use, Agents
- Last model added
- February 2026
- Scores refreshed
- November 20th, 2026
WildAWM + SFT tops the leaderboard with 31.9% Qwen3.5-4B leads with 22.3%
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 with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Qwen3.5-4B, at 22.3%. This view shows only models we track, each at its best.
- 1 WildAWM + SFT 31.9%
- 2 w/o Task Solv. 29.3%
- 3 w/o CF Probe 27.9%
- 4 w/o Task Diff. 26.4%
- 5 w/o Init. Unsat. 24.1%
- 6 w/o Judge Cons. 22.6%
- 7
Qwen3.5-4B 22.3%
Source: The WildAWM validation paper
- 1
Qwen3.5-4B 22.3%
Source: The WildAWM validation paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | WildAWM + SFT | 31.9% | Official leaderboard | ||
| 2 | w/o Task Solv. | 29.3% | Official leaderboard | ||
| 3 | w/o CF Probe | 27.9% | Official leaderboard | ||
| 4 | w/o Task Diff. | 26.4% | Official leaderboard | ||
| 5 | w/o Init. Unsat. | 24.1% | Official leaderboard | ||
| 6 | w/o Judge Cons. | 22.6% | Official leaderboard | ||
| 7 | | 22.3% | Feb 2026 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 22.3% | Feb 2026 | Official leaderboard |
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official WildAWM validation 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 Computer and browser use
- BrowseComp In use Hard-to-find facts that need persistent web browsing. Short verifiable answers graded for exact correctness. Claude Fable 5 93.3%
- ScreenSpot-Pro In use GUI grounding on high-resolution professional-software screenshots. The model must locate the target UI element. GPT-6 Astra 92.7%
- AndroidWorld In use Agents complete 116 tasks across 20 real Android apps in an emulator. Each task checks the device's system state to judge success. GPT-6 Astra 100%
- OSWorld 2.1 In use Agents complete 108 long-horizon workflows across 31 self-hosted websites and desktop apps, averaging 318 tool calls. Scored by binary completion and checkpoint partial credit. Claude Opus 5 44.3%
- WebArena-Verified In use Agents complete 812 audited WebArena tasks on self-hosted websites. Deterministic evaluators check the agent's answer and captured network traces.
Muse Spark 1.1 69%
- OSWorld 2.1 (offline subset) In use Agents complete the 82 OSWorld 2.1 tasks on the official offline list, with no internet access. Scored by checkpoint partial credit and strict pass rate. Claude Fable 5.2 89.3%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 9 months ago, in February 2026, for Qwen3.5-4B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (7 entries).
- Pass Models scored 1 model we track, plus 6 other entries on its leaderboard.
- Warning Newest score The newest came 9 months ago, in February 2026, for Qwen3.5-4B.
- Note Frontier models None of its models are on the capabilities index.
- Pass Headroom The best is 22.3%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 1 row for models we track and 6 other entries 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.