GPT-5.5
#12 of 274 scored.
OpenAIApr 23, 2026Closed
Capabilities Index
Range 156.7–162.3
159.1
Context
Takes file, image, text
1.05M
Input
Per 1M tokens
$5.00
Output
Per 1M tokens
$30
Benchmarks
How it scores.
- GPQA Diamond92.0%
10th of 186
- SWE-bench Verified80.6%
2nd of 32
- FrontierMath85.3%
12th of 81
- ARC-AGI-285.0%
9th of 79
- SimpleQA Verified63.0%
13th of 77
- Mock AIME100%
1st of 176
- Terminal-Bench84.7%
1st of 35
Everything else Epoch has for it
- ARC-AGI95.0%
- DTBench93.3%
- Surface Evolver Bench88.1%
- WeirdML84.9%
- FrontierMath-Tier-4-v2-Private72.5%
- DeepSWE67.0%
- LMCA63.8%
- SimpleBench62.8%
- APEX-Agents55.1%
- DeepResearch Bench54.0%
- Chess Puzzles51.6%
- Mystery Game Puzzles51.5%
- ProofBench50.0%
- ExploitBench47.4%
- FrontierCode43.0%
- GSO-Bench40.2%
- EBR-bench34.3%
- PostTrainBench27.2%
- CL-bench Life22.2%
- Furniture Assembly20.2%
- OSWorld 2.013.0%
- MirrorCode10.0%
- Remote Labor Index6.3%
On the index
Around it.
Sources