BeyondAIME — Published harder AIME-style extension math benchmark (ByteDance)
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Sarvam 105B | Sarvam AI | 69.1 | 1 | 2026-08-23 |
| 2 | Seed Oss 36B Instruct | ByteDance | 65 | 1 | 2026-06-15 |
| 3 | Sarvam 30B | Sarvam AI | 58.3 | 1 | 2026-08-23 |
| 4 | Qwen3 30B A3B | Alibaba | 56 | 1 | 2026-06-15 |
| 5 | Qwen3 32B | Alibaba | 29 | 1 | 2026-06-15 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.