GPT-5.5 Instant (OpenAI) 5 May 2026
GPT-5.5 Instant
GPT-5.5 Instant places in the top three on 2 benchmarks.
OpenAI Released 5 May 2026
27
Benchmarks with a score
0
Ranked first, out of five or more models
2
In the top three
OpenAI May 2026
GPT-5.5 Instant
Closed weights
142.5
#108 of 271
Epoch Capabilities Index
- MMMU-Pro 76% #29 /72
- HealthBench Professional 38.4% #24 /25
- CritPt 0% #158 /160
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,473 | #37 of 333 | Official leaderboard | |
| OpenAI user-flagged hallucination eval | 46.1% | #1 of 4 | Model card | |
| Document Arena | 1,460 | #22 of 37 | Official leaderboard | |
| Factuality Heavy prompts | 17.9% | #3 of 3 | Model card | |
| High Stakes factuality prompts | 19.9% | #3 of 3 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 82.5% | #81 of 219 | Epoch AI | |
| OTIS Mock AIME 2024-2025 | 68.1% | #104 of 202 | Epoch AI | |
| AIME 2025 | 81.2% | #87 of 143 | Model card | |
| FrontierMath Tiers 1-3 v2 | 26.3% | #71 of 89 | Epoch AI | |
| FrontierMath Tier 4 v2 | 2.44% | #68 of 71 | Epoch AI | |
| CritPt | 0% | #158 of 160 | Epoch AI |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| HealthBench Consensus | 94.7% | #13 of 17 | Model card | |
| HealthBench | 51.4% | #28 of 30 | Model card | |
| HealthBench Hard | 22.9% | #20 of 21 | Model card | |
| HealthBench Professional | 38.4% | #24 of 25 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vision Arena | 1,276 | #26 of 136 | Official leaderboard | |
| CharXiv | 81.6% | #23 of 67 | Model card | |
| MMMU-Pro | 76% | #29 of 72 | Model card |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Capture the Flag (Professional) | 94.1% | #2 of 21 | Model card | |
| First-Person Fairness evaluation | 1.01% | #2 of 10 | Model card | |
| CVE-Bench | 90.2% | #4 of 14 | Model card | |
| Cyber Range | 76.9% | #4 of 10 | Model card | |
| ProtocolQA Open-Ended | 36.7% | #12 of 25 | Model card | |
| Multimodal Troubleshooting Virology | 51.1% | #15 of 26 | Model card | |
| TroubleshootingBench | 33.7% | #12 of 19 | Model card | |
| Tacit Knowledge and Troubleshooting | 68.9% | #19 of 29 | Model card | |
| Agent JSK | 99.2% | #5 of 7 | Model card |
OpenAI's models over time
OpenAI's models in color, against every other model on the Epoch Capabilities Index, by release date.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it. 22 of them aren't in the launch post.
GPT-5.5 Instant System Card 4 May 2026
Health
Hallucination
Safety
Robustness
Bias
Bio and chem risk
Cyber risk
- Capture the Flag (Professional) 94.1
- CVE-Bench 90.2
- Cyber Range 76.9