Doubao-Seed-2.0-Lite-260215: benchmark results
Provider: ByteDance. Access: Open.
Unified ELO 1740 ± 21, rank #129 of 1645 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Reasoning - Common (CompassBench 2604) | 78.1 | Score (%) | 95.5 |
| OpenCompass Language - Creation (CompassBench 2604) | 77.1 | Score (%) | 88.6 |
| KINA | 41.49 | Accuracy (%) | 85.4 |
| SIGNPOST-Bench - Original Image Localization | 58.36 | Weighted localization accuracy (%; exp(-0.005 x geodesic km) | 84.2 |
| SIGNPOST-Bench - Text-Removed Localization | 47.2 | Weighted localization accuracy (%; exp(-0.005 x geodesic km) | 84.2 |
| ImmersedPrivacy - Social Context Rating Error (Audio as Text) | 1.22 | Mean absolute error (Tier 2: the model's 1-5 appropriateness | 81.8 |
| SIGNPOST-Bench - Random-Text Localization | 37.97 | Weighted localization accuracy (%; exp(-0.005 x geodesic km) | 78.9 |
| SIGNPOST-Bench - Similar-Text Localization | 53.09 | Weighted localization accuracy (%; exp(-0.005 x geodesic km) | 78.9 |
| OpenCompass LLM - Agent (CompassBench 2604) | 42.4 | Score (%) | 77.3 |
| OpenCompass Knowledge - Science (CompassBench 2604) | 91.7 | Score (%) | 75 |
| OpenCompass LLM - Language (CompassBench 2604) | 74.4 | Score (%) | 75 |
| COHERENCE - WikiHow | 72.74 | Exact-match accuracy (%) on the 2,076 WikiHow procedural doc | 72.2 |
Interactive version: theaggregate.ai/model?slug=doubao-seed-2-0-lite-260215 · How It Works · Data refreshed daily, snapshot 2026-10-10.