GPT-5.6 Sol (Max): benchmark results
GPT-5.6 Sol evaluated at the max reasoning-effort setting. Provider: OpenAI. Released 2026-07-09. Access: API.
Aggregate Rating 1886 ± 6, rank #42 of 2168 rated models, from 410 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Terminal-Bench Hard | 65.91 | Accuracy (%) | 100 |
| AlgoWorlds - Feasibility | 97.5 | Feasibility (%; share of worlds where the committed decision | 100 |
| AlgoWorlds - Reference Utility | 74.98 | Reference utility (%; x100 of a 0-1 score per feasible decis | 100 |
| BrowseComp (Mercor 130) | 95.7 | Pass@1 (%) | 100 |
| DramaChain Bench - Storyboard Design | 4.4 | Composite rubric score (1-5; mean of 8 leaf dimensions) | 100 |
| E-Commerce Bench | 1431 | Final Assets (¥k, mean of 5 episodes) | 100 |
| Epoch AI - Critpt | 32.29 | Score | 100 |
| Evo-Bench | 46.3 | Held-out score (%; the fixed DeepSeek-V4-Flash policy runs t | 100 |
| Forecast-Dojo (No Tools) | 0.65 | Multiclass Brier score (0-2 scale, lower is better; mean ove | 100 |
| Forecast-Dojo (Research Tools + Memory) | 0.55 | Multiclass Brier score (0-2 scale, lower is better; mean ove | 100 |
| Forecast-Dojo (Research Tools) | 0.55 | Multiclass Brier score (0-2 scale, lower is better; mean ove | 100 |
| HWE-Bench (Codex CLI) | 79.9 | Resolved (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-5-6-sol-max · How It Works · Data refreshed daily, snapshot 2026-10-20.