Spacebar Workspace Pilot - Reliability and recovery: leaderboard
Spacebar October14 workspace pilot:14sequential user requests in a persistent workspace, scored on retained artifacts; one campaign per model. Direct/OpenRouter route and requested medium/max/high setting follow source; no repeated-run estimate. Research,spreadsheet,collaborative app,memo,consistency and recovery rubric. Rubric Points.
Metric: Rubric Points. Source: www.spacebar.ai. Saturation forecast: Around June 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6.1 Sol (Medium) | 11 | #7 (GPT-6.1 Sol) |
| 2 | Claude Opus 5.5 (Medium) | 11 | #4 (Claude Opus 5.5) |
| 3 | DeepSeek V4.1 Flash (Max) | 10 | #35 (DeepSeek V4.1 Flash) |
| 4 | GLM-5.3 (Max) | 8 | #37 (GLM-5.3) |
| 5 | Grok 4.7 (High) | 8 | #27 (Grok 4.7) |
| 6 | Muse Spark 1.3 (High) | 8 | #25 (Muse Spark 1.3) |
No result here: #6 GPT-6 Astra, #8 Gemini 4 Argon, #9 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=spacebar-workspace-pilot-reliability-and-recovery · How It Works · Data refreshed daily, snapshot 2026-10-22.