Doubao-Seed-2.0-Pro-260215: benchmark results
ByteDance's Doubao-Seed-2.0-Pro (February 2026), the flagship general-purpose model of the Seed 2.0 family for complex reasoning and long-chain agent tasks. Provider: ByteDance. Released 2026-02-15. Access: API.
Aggregate Rating 1784 ± 11, rank #77 of 1656 rated models, from 122 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ChronoPhyBench - Frame Sorting (3 Frames) | 74 | Accuracy (%) on ChronoPhyBench chronological sorting of 3 he | 100 |
| DiscoBench - Neutral - Accuracy | 43.1 | End-to-end accuracy (%) | 100 |
| OpenCompass Knowledge - Humanities (CompassBench 2604) | 95 | Score (%) | 100 |
| OpenCompass LLM - Math (CompassBench 2604) | 77.3 | Score (%) | 100 |
| OpenCompass LLM - Reasoning (CompassBench 2604) | 65.2 | Score (%) | 100 |
| OpenCompass Language - Chat (CompassBench 2604) | 91.7 | Score (%) | 100 |
| OpenCompass Math - College (CompassBench 2604) | 83.8 | Score (%) | 100 |
| OpenCompass Reasoning - Common (CompassBench 2604) | 82.1 | Score (%) | 100 |
| SuperCLUE General (2025 Annual) - Math Reasoning | 87.83 | Score | 100 |
| SuperCLUE-Fin (March 2026) - Overall | 79.8 | Score | 100 |
| SuperCLUE-VLM (April 2026) - Overall | 90.66 | Score | 100 |
| Video-MME-v2 | 53.1 | Avg Accuracy w/o sub (%) | 97.9 |
Interactive version: theaggregate.ai/model?slug=doubao-seed-2-0-pro-260215 · How It Works · Data refreshed daily, snapshot 2026-10-20.