Benchmark Almanac

Seed2.1 Pro

Seed2.1 Pro ranks first on 16 of the 72 benchmarks it has scores on.

ByteDance Seed Released 23 Jun 2026 Family Seed 2 The lab's model card
101
Benchmarks with a score
16
Ranked first, out of five or more models
25
In the top three
ByteDance Seed Jun 2026

Seed2.1 Pro

Closed weights Flagship

No Epoch Capabilities Index score yet
  • MMMU-Pro 82.7% #3 /72
  • SciCode 59.8% #1 /47
  • NL2Repo-Bench 47% #15 /28

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
Humanity's Last Exam 55.7% #21 of 147 Model card
SimpleVQA 74.5% #4 of 23 Model card
SuperGPQA 70.8% #5 of 26 Model card
MMLongBench 78.3% #1 of 4 Model card
Video SimpleQA 76.4% #1 of 4 Model card
HLE-Verified 42.9% #4 of 4 Model card
Doubao Multi-Turn Bench 52.5% #1 of 1 Model card
KINA 48.3% #1 of 1 Model card
MSQA 50.2% #1 of 1 Model card
xDailyBench 61% #1 of 1 Model card

Math and science

BenchmarkScoreRankSource
Beyond AIME 87% #1 of 14 Model card
FrontierScience 75% #2 of 7 Model card
ARC-AGI-2 62.5% #10 of 22 Model card
FrontierScience-Olympiad 75% #1 of 2 Model card
xbench-ScienceQA 71% #15 of 29 Official leaderboard
FrontierScience-Research 28.3% #2 of 3 Model card
MathArena (Apex, IMO, USAMO) 31.3% #10 of 13 Model card
IMO-ProofBench 54.3% #4 of 5 Model card
IMO 2025 65.2% #5 of 6 Model card
ExplorationBench 40% #9 of 10 Official leaderboard
HorizonMath 2.00% #3 of 3 Model card
IPhO 2025 79.3% #1 of 1 Model card
LiveMathematicianBench 20.9% #1 of 1 Model card

Coding

BenchmarkScoreRankSource
SciCode 59.8% #1 of 47 Model card
AetherCode 65.8% #1 of 3 Model card
Code Arena 1,519 #50 of 124 Official leaderboard
SWE-Bench Pro 57.5% #35 of 77 Model card
Monthly-SWEBench 53% #6 of 12 Official leaderboard
NL2Repo-Bench 47% #15 of 28 Model card
ProgramBench 50.3% #13 of 20 Model card
Image-to-WebDev Arena 1,510 #37 of 55 Official leaderboard
EvoCode-Bench v2 6.70% #11 of 13 Official leaderboard
SWE Atlas 35.2% #10 of 11 Model card
DeepSWE 32.7% #49 of 51 Model card
PostTrainBench 16.5% #7 of 7 Model card

Agents and tools

BenchmarkScoreRankSource
OSWorld 78.8% #1 of 12 Model card
MCP Atlas 83.8% #6 of 40 Model card
BrowseComp 86.2% #16 of 77 Model card
SkillsBench 60.4% #4 of 13 Model card
WildClawBench 61.7% #2 of 5 Model card
Toolathlon 50.6% #19 of 37 Model card
Terminal-Bench 2.1 71% #34 of 53 Model card
MobileWorld 73.1% #2 of 3 Model card
Long-Horizon-Terminal-Bench 2.0 46.4% #11 of 15 Official leaderboard
Long-Horizon-Terminal-Bench (LHTB) 28.6% #19 of 25 Official leaderboard
CreativeWork 42.5% #1 of 1 Model card
GameWorld 31.2% #1 of 1 Model card
SeedClawBench 66.6% #1 of 1 Model card

Professional work

BenchmarkScoreRankSource
GDPval 87.9% #1 of 10 Model card
OfficeQA Pro 70.9% #2 of 14 Model card
APEX-Agents 33.8% #6 of 12 Model card
$OneMillion-Bench 23.7% #8 of 13 Official leaderboard
Vals Finance Agent 60.7% #5 of 8 Model card
BlenderWorld 51.3 #9 of 13 Official leaderboard
Agents' Last Exam 19.5% #25 of 25 Model card
PresentBench 54.6% #3 of 3 Model card
WorkSpaceBench 53% #3 of 3 Model card
Agent Startup Bench 68.8% #1 of 1 Model card

Vision, audio and robotics

BenchmarkScoreRankSource
MathVista 90.7% #1 of 64 Model card
BLINK 81.4% #1 of 28 Model card
MMMU-Pro 82.7% #3 of 72 Model card
MathVerse 89.7% #1 of 12 Model card
LongVideoBench 80.6% #1 of 11 Model card
Video-MME 89.2% #4 of 42 Model card
MATH-Vision (MathVision) 94.5% #5 of 42 Model card
CharXiv (DQ) 95.5% #1 of 7 Model card
TreeBench 71.1% #1 of 7 Model card
VisuLogic 54.3% #1 of 7 Model card
ERQA 72% #4 of 26 Model card
RealWorldQA 86.7% #5 of 30 Model card
MINERVA 70.7% #1 of 6 Model card
Video-Holmes 68.2% #1 of 6 Model card
OVBench 70% #1 of 5 Model card
OVO-Bench 80.7% #1 of 5 Model card
TOMATO 79.5% #1 of 5 Model card
MotionBench 74.9% #2 of 9 Model card
CharXiv 86.4% #15 of 67 Model card
LVBench 78% #6 of 25 Model card
DUDE 82.8% #1 of 4 Model card
TVBench 80.5% #2 of 7 Model card
BabyVision 73.7% #8 of 27 Official leaderboard
CrossVid 65% #1 of 3 Model card
EMMA 79.3% #1 of 3 Model card
MMSI-Bench 35.9% #2 of 5 Model card
VLMsAreBiased 83.6% #2 of 5 Model card
WorldVQA 53% #4 of 9 Model card
OCRBench v2 63.2% #9 of 20 Model card
DynaMath 73.1% #8 of 17 Model card
ChartQA Pro 70.9% #4 of 8 Model card
VisFactor 51.4% #2 of 4 Model card
ZeroBench 22% #16 of 25 Model card
ContPhy 63.6% #2 of 3 Model card
EmbSpatialBench 83.4% #7 of 10 Model card
Claw-Eval-MM 51% #6 of 7 Model card
LiveSports-3K 76.8% #4 of 4 Model card
Image2FloorPlan 48.0 #1 of 1 Model card
MeasureBench 62.9% #1 of 1 Model card
WorldBench 67.6% #1 of 1 Model card

Safety and security

BenchmarkScoreRankSource
SUPERChem 59.8% #2 of 5 Model card
CyberGym 70.2% #23 of 33 Model card

Other

BenchmarkScoreRankSource
SuperCLUE 65.9 #9 of 18 Official leaderboard

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Its system card

Every result in the longer documents the lab published with it. 54 of them aren't in the launch post.

Seed2.1 Model Card 23 Jun 2026

Capabilities

Cyber risk

AI R&D risk