Ring-flash-2.0: benchmark results
InclusionAI's open 100B MoE thinking model (6.1B active), the RL-trained reasoning counterpart of Ling-flash-2.0 (September 2025). Provider: InclusionAI. Released 2025-09-19. Access: Open.
Aggregate Rating 1598 ± 32, rank #446 of 1656 rated models, from 79 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SuiChat-CN - Few-Shot | 76 | Macro F1 (%) of suicide-risk level classification of Chinese | 95.3 |
| AA AIME 2025 | 83.67 | Accuracy (%) | 80.4 |
| AA LiveCodeBench | 62.75 | Pass@1 (%) | 68.7 |
| AA MMLU-Pro | 79.27 | Accuracy (%) | 63.5 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 20 | Accuracy (%) | 63 |
| AA Omniscience - Software Engineering (SWE) - Swift | 36 | Accuracy (%) | 60.6 |
| SuiChat-CN | 68.7 | Standard F1 (self-reported) | 58.1 |
| AA Omniscience - Software Engineering (SWE) - Julia | 12 | Accuracy (%) | 58 |
| AA Omniscience - Software Engineering (SWE) - Python | 19.5 | Accuracy (%) | 56.1 |
| AA GPQA Diamond | 72.53 | Accuracy (%) | 51.9 |
| SuiChat-CN - Without Keywords | 66.4 | Macro F1 (%) of suicide-risk level classification of Chinese | 51.2 |
| AA Omniscience - Software Engineering (SWE) - C | 31 | Accuracy (%) | 50.5 |
Interactive version: theaggregate.ai/model?slug=ring-flash-2-0 · How It Works · Data refreshed daily, snapshot 2026-10-20.