GPT-5.1 (Medium): benchmark results
GPT-5.1 evaluated at the medium reasoning-effort setting. Provider: OpenAI. Released 2025-11-12. Access: API.
Aggregate Rating 1707 ± 11, rank #337 of 2168 rated models, from 163 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| FireBench - Item Ranking | 93 | Share of the 200 item ranking samples (%) in which the model | 100 |
| HardSecBench - Refinement@5 (GPT-5.1 Collaborator) - Security | 53.88 | Requirement Pass Rate (%) | 100 |
| Medmarks - CareQA Open | 89.02 | Score (%) | 100 |
| Medmarks - MEDEC | 70.46 | Score (%) | 100 |
| Medmarks - SCTpublic | 77.96 | Score (%) | 100 |
| NLCO - Set-L - Feasibility Rate | 95.5 | Average feasibility rate: output satisfies every constraint | 100 |
| NLCO - Set-M - Feasibility Rate | 98 | Average feasibility rate: output satisfies every constraint | 100 |
| Medmarks - HEAD-QA v2 | 92.33 | Score (%) | 98.9 |
| Medmarks - MedConceptsQA Easy | 99.97 | Score (%) | 98.9 |
| Medmarks - MedMCQA | 84.65 | Score (%) | 98.9 |
| Medmarks - MetaMedQA | 82.74 | Score (%) | 98.9 |
| Medmarks - SuperGPQA Medicine Easy | 68.87 | Score (%) | 98.9 |
Interactive version: theaggregate.ai/model?slug=gpt-5-1-medium · How It Works · Data refreshed daily, snapshot 2026-10-20.