Seed 2.1 Pro Preview
Seed 2.1 Pro Preview is at the frontier in multimodal tasks, capable in agentic tasks, and behind the leaders in coding and reasoning. Too few results yet to rate safety, long context, math, multilingual tasks, instruction following, or factuality.
Capability profile
Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.
Too few results yet to rate Safety, Long Context, Math, Multilingual, Instruction Following or Factuality.
Price
No current price is tracked for this model. See the rate card
Evidence
66results on63benchmarks
- 1 independently verified
- 65 vendor-reported
From 4 sources · latest Oct 7, 2026 · How verification works
Seed 2.1 Pro Preview benchmark results
66 results on 63 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.
2.7% behind the leader4 of 6 ranked benchmarks measured
- 90.70Oct 7, 2026
- 89.20Oct 7, 2026
- 82.70Oct 7, 2026
- CharXiv (reasoning)86.40Oct 7, 2026CharXiv-R
Show 3 more multimodal resultsHide 3 multimodal results
- 81.40Oct 7, 2026
- 81.60Sep 28, 2026
- 63.20Oct 7, 2026
21.4% behind the leader4 of 7 ranked benchmarks measured
- GDPval (win rate)87.90Sep 28, 2026GDPval
- 83.80Oct 7, 2026
- 86.20Oct 7, 2026
- OSWorld-Verified78.80Oct 7, 2026OSWorld
Show 1 more agentic resultHide 1 agentic result
- AA ApexAgents33.80Oct 7, 2026APEX-Agents
26.4% behind the leader4 of 10 ranked benchmarks measured
- 59.80Oct 7, 2026
- 71.00Oct 7, 2026
- 1518.86Jun 24, 2026
- 57.50Oct 7, 2026
29.7% behind the leader2 of 6 ranked benchmarks measured
- 55.70Oct 7, 2026
- 62.50Oct 7, 2026
More results
Benchmarks outside the capability baskets. They are not ranked against a leader.
- MathArena Apex (Pass@1)31.30Oct 7, 2026MathArena Apex
- 80.60Oct 7, 2026
- 74.90Oct 7, 2026
- 80.50Oct 7, 2026
- 68.20Oct 7, 2026
- 82.80Oct 7, 2026
Show 42 more resultsHide 42 results
- 41.40Oct 7, 2026
- 51.00Oct 7, 2026
- 73.70Oct 7, 2026
- 87.00Oct 7, 2026
- BrowseComp (with Search)86.20Sep 28, 2026
- 70.90Oct 7, 2026
- 70.20Oct 7, 2026
- 32.70Oct 7, 2026
- 73.10Oct 7, 2026
- 83.40Oct 7, 2026
- 72.00Oct 7, 2026
- Finance Agent v1.160.70Oct 7, 2026
- 55.70Sep 28, 2026
- HLE-Verified (no tool)42.90Sep 28, 2026
- 78.00Oct 7, 2026
- 94.50Oct 7, 2026
- 89.70Oct 7, 2026
- MathVerse (Vision-Only)89.70Sep 28, 2026
- 70.70Oct 7, 2026
- 78.30Oct 7, 2026
- 35.90Oct 7, 2026
- MMSIBench (circular)35.90Sep 28, 2026
- 50.20Oct 7, 2026
- 47.00Oct 7, 2026
- Office QA Pro [Multimodal]72.20Sep 28, 2026
- 72.20Oct 7, 2026
- 70.90Sep 28, 2026
- 70.00Oct 7, 2026
- 16.50Oct 7, 2026
- 50.30Oct 7, 2026
- 86.70Oct 7, 2026
- 74.50Oct 7, 2026
- 70.80Oct 7, 2026
- 79.50Oct 7, 2026
- 50.60Oct 7, 2026
- 76.40Oct 7, 2026
- 54.30Sep 28, 2026
- 0.54Sep 28, 2026
- 53.00Oct 7, 2026
- 18.00Oct 7, 2026
- ZeroBench (main)18.00Sep 28, 2026
- ZeroBench (sub)49.40Sep 28, 2026
Seed 2.1 Pro Preview: common questions
Who makes Seed 2.1 Pro Preview?
Seed 2.1 Pro Preview is made by ByteDance.
What is Seed 2.1 Pro Preview good at?
Seed 2.1 Pro Preview is at the frontier in multimodal tasks, capable in agentic tasks, and behind the leaders in coding and reasoning. Too few results yet to rate safety, long context, math, multilingual tasks, instruction following, or factuality.
How many benchmarks has Seed 2.1 Pro Preview been tested on?
We track 66 results for Seed 2.1 Pro Preview on 63 benchmarks from 4 sources, 1 of them independently verified. The latest was recorded on Oct 7, 2026.
About this record
Where Seed 2.1 Pro Preview's numbers come from, and every name it appears under.
- Tracked since
- Jun 24, 2026
- Newest source mention
- Sep 26, 2026
Where the results come from
Verification: 66 scores · 1 independently verified · 65 vendor-reported. How these tiers are assigned
From 4 sources on 3 sites. api.llm-stats.com supplies 53 of them; the 1 independently verified result comes from 1 site. Bars are coloured by trust tier.
- api.llm-stats.com53
- lf3-static.bytednsdoc.com12
- datasets-server.huggingface.co1