Seed 2.0 Pro: benchmark results
ByteDance's flagship Seed 2.0 Pro reasoning and agent model. Provider: ByteDance. Released 2026-02-14. Access: API.
Aggregate Rating 1751 ± 4, rank #118 of 1656 rated models, from 587 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AGI-Eval Community - Subject Knowledge | 91.1 | Accuracy (%) | 100 |
| AeroGround | 54.4 | Accuracy (%; all 2,250 aerial-ground collaborative questions | 100 |
| AeroGround - Multi-Target Matching | 30.6 | Accuracy (%; Object Alignment: give the full one-to-one corr | 100 |
| AeroGround - Multi-Target Relation | 83.3 | Accuracy (%; Spatial Understanding: order comparison targets | 100 |
| AeroGround - Occluded Target Judgment | 45 | Accuracy (%; Spatial Reasoning: decide whether a target is v | 100 |
| AeroGround - Relative Orientation Determination | 43.3 | Accuracy (%; Spatial Understanding: select the egocentric qu | 100 |
| AeroGround - Reverse Sequence Reasoning | 77.5 | Accuracy (%; Spatial Reasoning: find the ground frame that d | 100 |
| AeroGround - Single-Target Matching | 50 | Accuracy (%; Object Alignment: select the candidate in the o | 100 |
| AgroTools - Argument Accuracy | 49.34 | Argument accuracy (%): share of gold tool-call steps where t | 100 |
| AgroTools - Tool Selection Accuracy | 95.16 | Tool accuracy (%): share of gold tool-call steps where the m | 100 |
| AirGroundBench | 57.14 | Average accuracy (%) over the nine multiple-choice VQA tasks | 100 |
| AirGroundBench - Cross-View Matching | 56.83 | Accuracy (%) on the cross-view matching VQA task; higher is | 100 |
Interactive version: theaggregate.ai/model?slug=seed-2-0-pro · How It Works · Data refreshed daily, snapshot 2026-10-20.