Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
coding
| Benchmark | OpenAI Codex | Step3-VL-10B |
|---|---|---|
| SWE-bench Verified | 69.1 | — |
multimodal
| Benchmark | OpenAI Codex | Step3-VL-10B |
|---|---|---|
| MMMU | — | 78.1 |
math
| Benchmark | OpenAI Codex | Step3-VL-10B |
|---|---|---|
| AIME 2025 | — | 87.7 |
Category Winners
coding
59.1
math
S
Step3-VL-10B65.9
multimodal
S
Step3-VL-10B77.2
agent
78.2