Seed 2.0 Mini: benchmark results
ByteDance Seed2.0 Mini is the throughput-oriented model in the Seed2.0 API family. Source: https://seed.bytedance.com/en/blog/seed-2-0-official-launch Provider: ByteDance. Released 2026-02-14. Access: API.
Unified ELO 1721 ± 25, rank #287 of 2656 rated models, from 79 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLM Stats (Graphwalks parents <128k) | 93 | Score (%) | 100 |
| Polish Cultural Vision Benchmark - Country | 51.82 | Accuracy (%) | 92.7 |
| LLM Stats (CharXiv-D) | 91.9 | Score (%) | 88.2 |
| Polish Cultural Vision Benchmark - Object | 27.8 | Accuracy (%) | 85.4 |
| LLM Stats (BLINK) | 73.4 | Score (%) | 81.2 |
| LLM Stats (ZEROBench-Sub) | 36.2 | Score (%) | 80 |
| FinRiskAtlas - Evidence-Grounded Processing - Quantitative Reasoning | 79.31 | Numerical Accuracy (%) | 79.7 |
| RealWorldQA | 81.6 | RealWorldQA (self-reported) | 78.1 |
| FinRiskAtlas - Applied Review - Disputed-Issue Generation | 72.73 | Rubric Score (%; DeepSeek-V4-Flash judge) | 77.4 |
| LLM Stats (MuirBench) | 78 | Score (%) | 76.9 |
| AI Chess Leaderboard (Continuation) | 832 | Elo | 73.3 |
| LLM Stats (MMStar) | 79.1 | Score (%) | 73.1 |
Interactive version: theaggregate.ai/model?slug=seed-2-0-mini · How It Works · Data refreshed daily, snapshot 2026-09-19.