ByteDance model product
ByteDance's flagship next-generation agent model built for real-world productivity. A deep-thinking model with strong demand understanding, long-horizon planning, and continuous self-repair, it delivers reliable end-to-end results across complex coding, long-chain agents, and multi-step engineering workflows. Seed 2.1 Pro also advances knowledge, reasoning, and multimodal understanding, with SOTA results across several video understanding benchmarks. Served via Volcano Engine as Doubao-Seed-2.1-pro.
Updated Aug 10, 2026. Default version: Seed 2.1 Pro
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version Seed 2.1 Pro.
| Agent Startup Bench | 0.7 | 1 | 2 | 100.0% | C | |
| ArcAGI2 | 0.6 | 1 | 2 | 100.0% | C | |
| Artifacts Bench | 0.5 | 1 | 3 | 100.0% | C | |
| BLINK | 0.8 | 1 | 13 | 100.0% | C | |
| ChartQAPro | 0.7 | 1 | 2 | 100.0% | C | |
| CharXiv-D | 1.0 | 1 | 16 | 100.0% | C | |
| ContPhy | 0.6 | 1 | 2 | 100.0% | C | |
| CreativeWork | 0.4 | 1 | 2 | 100.0% | C | |
| CrossVid | 0.7 | 1 | 2 | 100.0% | C | |
| Doubao Multi-Turn Bench | 0.5 | 1 | 2 | 100.0% | C | |
| EMMA | 0.8 | 1 | 2 | 100.0% | C | |
| Finance Agent v1.1 | 0.6 | 1 | 2 | 100.0% | C | |
| GameWorld | 0.3 | 1 | 2 | 100.0% | C | |
| GDPval | 0.9 | 1 | 3 | 100.0% | C | |
| Image2FloorPlan | 0.5 | 1 | 2 | 100.0% | C | |
| IMOProof-Adv | 0.5 | 1 | 1 | 100.0% | C | |
| KINA | 0.5 | 1 | 2 | 100.0% | C | |
| LongVideoBench | 0.8 | 1 | 4 | 100.0% | C | |
| MathVerse | 0.9 | 1 | 2 | 100.0% | C | |
| MathVista | 0.9 | 1 | 39 | 100.0% | C | |
| MeasureBench | 0.6 | 1 | 2 | 100.0% | C | |
| Minerva | 0.7 | 1 | 2 | 100.0% | C | |
| MMLongBench-128K | 0.8 | 1 | 2 | 100.0% | C | |
| MMSIBench | 0.4 | 1 | 2 | 100.0% | C | |
| MotionBench | 0.7 | 1 | 3 | 100.0% | C | |
| MSQA | 0.5 | 1 | 2 | 100.0% | C | |
| OfficeQA Pro | 0.7 | 1 | 7 | 100.0% | C | |
| OneMillion Bench | 0.7 | 1 | 2 | 100.0% | C | |
| OSWorld | 0.8 | 1 | 20 | 100.0% | C | |
| OVBench | 0.7 | 1 | 2 | 100.0% | C | |
| OVOBench | 0.8 | 1 | 2 | 100.0% | C | |
| PresentBench | 0.5 | 1 | 2 | 100.0% | C | |
| Repo Env | 0.6 | 1 | 2 | 100.0% | C | |
| SciCode | 0.6 | 1 | 18 | 100.0% | C | |
| SeedClawBench | 0.7 | 1 | 2 | 100.0% | C | |
| SuperChem | 0.6 | 1 | 3 | 100.0% | C | |
| SWE-Atlas | 0.4 | 1 | 2 | 100.0% | C | |
| TOMATO | 0.8 | 1 | 2 | 100.0% | C | |
| Trae Code Gen | 0.6 | 1 | 2 | 100.0% | C | |
| Trae Error Fix | 0.6 | 1 | 2 | 100.0% | C | |
| TreeBench | 0.7 | 1 | 2 | 100.0% | C | |
| TVBench | 0.8 | 1 | 3 | 100.0% | C | |
| VideoHolmes | 0.7 | 1 | 3 | 100.0% | C | |
| Video-MME | 0.9 | 1 | 17 | 100.0% | C | |
| VideoSimpleQA | 0.8 | 1 | 2 | 100.0% | C | |
| VisFactor | 0.5 | 1 | 3 | 100.0% | C | |
| VisuLogic | 0.5 | 1 | 3 | 100.0% | C | |
| VLMsAreBiased | 0.8 | 1 | 2 | 100.0% | C | |
| Web Bench | 0.8 | 1 | 2 | 100.0% | C | |
| WorldBench | 0.7 | 1 | 2 | 100.0% | C | |
| xDailyBench | 0.6 | 1 | 2 | 100.0% | C | |
| AetherCode | 0.7 | 2 | 2 | 0.0% | C | |
| APEX-Agents | 0.3 | 2 | 7 | 83.3% | C | |
| Beyond AIME | 0.9 | 2 | 5 | 75.0% | C | |
| ClawEval-MM | 0.5 | 2 | 3 | 50.0% | C | |
| DUDE | 0.8 | 2 | 2 | 0.0% | C | |
| ERQA | 0.7 | 2 | 23 | 95.5% | C | |
| FrontierCS | 0.5 | 2 | 2 | 0.0% | C | |
| FrontierScience Olympiad | 0.8 | 2 | 3 | 50.0% | C | |
| HorizonMath | 0.0 | 2 | 3 | 50.0% | C | |
| IMO 2025 | 0.7 | 2 | 2 | 0.0% | C | |
| IPhO 2025 | 0.8 | 2 | 2 | 0.0% | C | |
| LiveMathematicianBench | 0.2 | 2 | 2 | 0.0% | C | |
| LiveSports-3K | 0.8 | 2 | 2 | 0.0% | C | |
| LVBench | 0.8 | 2 | 24 | 95.7% | C | |
| MathVision | 0.9 | 2 | 32 | 96.8% | C | |
| MobileWorld | 0.7 | 2 | 3 | 50.0% | C | |
| WildClawBench | 0.6 | 2 | 4 | 66.7% | C | |
| Workspace Bench | 0.5 | 2 | 2 | 0.0% | C | |
| WorldVQA | 0.5 | 2 | 5 | 75.0% | C | |
| BabyVision | 0.7 | 3 | 9 | 75.0% | C | |
| FrontierScience Research | 0.3 | 3 | 4 | 33.3% | C | |
| MCP Atlas | 0.8 | 3 | 30 | 93.1% | C | |
| OCRBench_V2 | 0.6 | 3 | 7 | 66.7% | C | |
| RealWorldQA | 0.9 | 3 | 26 | 92.0% | C | |
| SimpleVQA | 0.7 | 3 | 13 | 83.3% | C | |
| ZEROBench | 0.2 | 3 | 9 | 75.0% | C | |
| Agents' Last Exam | 0.4 | 4 | 5 | 25.0% | C | |
| MMMU-Pro | 0.8 | 4 | 65 | 95.3% | C | |
| Program Bench | 0.5 | 4 | 5 | 25.0% | C | |
| SuperGPQA | 0.7 | 4 | 34 | 90.9% | C | |
| NL2Repo | 0.5 | 5 | 14 | 69.2% | C | |
| PostTrainBench | 0.2 | 5 | 5 | 0.0% | C | |
| DynaMath | 0.7 | 6 | 7 | 16.7% | C | |
| EmbSpatialBench | 0.8 | 6 | 8 | 28.6% | C | |
| MathArena Apex | 0.3 | 6 | 7 | 16.7% | C | |
| BrowseComp | 0.9 | 8 | 58 | 87.7% | C | |
| CyberGym | 0.7 | 8 | 11 | 30.0% | C | |
| DeepSWE | 0.3 | 8 | 10 | 22.2% | C | |
| CharXiv-R | 0.9 | 10 | 47 | 80.4% | C | |
| Humanity's Last Exam | 0.6 | 10 | 92 | 90.1% | C | |
| Terminal-Bench 2.1 | 0.7 | 13 | 17 | 25.0% | C | |
| Toolathlon | 0.5 | 13 | 31 | 60.0% | C | |
| SWE-Bench Pro | 0.6 | 22 | 44 | 51.2% | C |
Preference and agent-evaluation signals from published Arena datasets.
The default version has no published Arena rows, or its source alias has not been resolved.
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
The default version has no provider offering with current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| Seed 2.1 Pro | 87.4 | N/A | N/A | N/A | No | Proprietary |
A concise description based on the published model registry.
ByteDance's flagship next-generation agent model built for real-world productivity. A deep-thinking model with strong demand understanding, long-horizon planning, and continuous self-repair, it delivers reliable end-to-end results across complex coding, long-chain agents, and multi-step engineering workflows. Seed 2.1 Pro also advances knowledge, reasoning, and multimodal understanding, with SOTA results across several video understanding benchmarks. Served via Volcano Engine as Doubao-Seed-2.1-pro.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about Seed 2.1 Pro.
Seed 2.1 Pro's default version was released on Jun 24, 2026.
No official standard PAYG price is currently available for Seed 2.1 Pro.
Seed 2.1 Pro is published under ByteDance in the model registry.
The current registry does not publish a context window for the default version.
No. The default version is not marked as having publicly available weights.
No published provider offering is currently linked to the default version.
Nearby ranked alternatives include GPT-5.6-Sol, Qwen3.7 Max, GPT-5.6-Terra.