LongBench v2 — Context-length subset x no-CoT.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Qwen2.5 Instruct 72B | Alibaba | 44.4 | 1 | 2026-06-06 |
| 2 | MiniMax Text 01 | MiniMax | 43.5 | 1 | 2026-06-06 |
| 3 | GPT-4o | OpenAI | 40.2 | 1 | 2026-06-12 |
| 4 | Claude 3.5 Sonnet | Anthropic | 37 | 1 | 2026-06-12 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.