LongBench v2 — Aggregate across subsets, no-CoT axis.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | MiniMax Text 01 | MiniMax | 52.9 | 1 | 2026-06-06 |
| 2 | GPT-4o | OpenAI | 50.1 | 1 | 2026-06-06 |
| 3 | DeepSeek-V3 | DeepSeek | 48.7 | 1 | 2026-06-06 |
| 4 | Qwen2.5 Instruct 72B | Alibaba | 42.1 | 1 | 2026-06-06 |
| 5 | Claude 3.5 Sonnet | Anthropic | 41 | 1 | 2026-06-06 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.