NEO-2B-SFT: benchmark results
Provider: Other.
Aggregate Rating 1409 ± 44, rank #5680 of 10916 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Game2Real-Bench - Real World - Fine-grained Perception | 28 | Task score (%) | 18.8 |
| Game2Real-Bench - Real World - Spatial Reasoning | 21 | Task score (%) | 14.6 |
| VISTA-Bench (Pure Text) | 51.5 | Accuracy (%) over the same 1,500 questions given as ordinary | 13.8 |
| VISTA-Bench - Unimodal Knowledge | 29.4 | Accuracy (%) on the 500 unimodal knowledge questions (natura | 13.8 |
| VISTA-Bench - Multimodal Reasoning | 31.3 | Accuracy (%) on the 300 multimodal reasoning questions (logi | 12.1 |
| VISTA-Bench | 30.9 | Accuracy (%) over all 1,500 questions (300 perception, 300 r | 10.3 |
| Game2Real-Bench - Overall | 13.2 | Mean per-sample score (%) | 8.3 |
| Game2Real-Bench - Toy Game - Constraint Reasoning | 9.5 | Accuracy (%) | 8.3 |
| Game2Real-Bench - Toy Game - Sequential Planning | 10.2 | Accuracy (%) | 8.3 |
| VISTA-Bench - Multimodal Knowledge | 25.5 | Accuracy (%) on the 400 multimodal knowledge questions (STEM | 6.9 |
| VISTA-Bench - Multimodal Perception | 40 | Accuracy (%) on the 300 multimodal perception questions (glo | 6.9 |
| Game2Real-Bench - Real World - Constraint Reasoning | 15.1 | Task score (%) | 6.2 |
Interactive version: theaggregate.ai/model?slug=neo-2b-sft · How It Works · Data refreshed daily, snapshot 2026-11-08.