Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.1 Instant | GPT-5.2 Pro |
|---|---|---|
| GPQA Diamond | 88.1 | 93.2 |
| Humanity's Last Exam | — | 36.6 |
| ARC-AGI v2 | — | 54.2 |
math
| Benchmark | GPT-5.1 Instant | GPT-5.2 Pro |
|---|---|---|
| AIME 2025 | 94 | 100 |
| FrontierMath | 26.7 | — |
coding
| Benchmark | GPT-5.1 Instant | GPT-5.2 Pro |
|---|---|---|
| SWE-bench Verified | 76.3 | — |
multimodal
| Benchmark | GPT-5.1 Instant | GPT-5.2 Pro |
|---|---|---|
| MMMU | 85.4 | — |
agent
| Benchmark | GPT-5.1 Instant | GPT-5.2 Pro |
|---|---|---|
| BrowseComp | — | 77.9 |
Category Winners
coding
74.5
math
84.0
reasoning
85.8
multimodal
89.6
agent
74.4