Seed2.0 Pro

Seed2.0 Pro ranks first on 10 of the 129 benchmarks with five or more models scored.

ByteDance Seed Released February 14th, 2026 Family Seed 2 The lab's model card Compare with others

188

Benchmarks with a score

Ranked first, out of five or more models

In the top three

ByteDance Seed Feb 2026

Seed2.0 Pro

Closed weights Flagship

149.9 #94 of 404

Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.

MMMU-Pro 78.2% #26 /90
SciCode 48.5% #13 /49
BFCL V4 73.4% #2 /30

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

Benchmark	Score	Rank	Source
MMLU-Pro	87%	#11 of 167	Lab model card
LMArena (Elo)	1,457	#53 of 335	Official leaderboard
MultiChallenge	68.3%	#7 of 42	Lab model card
LongBench v2	63.8%	#6 of 35	Lab model card
Humanity's Last Exam	54.2%	#28 of 154	Lab model card
Global PIQA	92.3%	#3 of 14	Lab model card
HallusionBench	68%	#7 of 31	Lab model card
SuperGPQA	68.7%	#8 of 35	Lab model card
HLE-Verified	73.6%	#1 of 4	Lab model card
SimpleVQA	71.4%	#8 of 30	Lab model card
MMMLU	88.1%	#20 of 71	Lab model card
GraphWalks	83.3%	#9 of 27	Lab model card
Visual Aesthetic Benchmark	33.3%	#6 of 18	Official leaderboard
FRAMES	84.5%	#3 of 9	Lab model card
WideSearch	74.7%	#7 of 19	Lab model card
COLLIE	93.9%	#5 of 12	Lab model card
Seal-0	49.5%	#3 of 7	Lab model card
CL-bench Life	15.5%	#7 of 16	Official leaderboard
CL-bench	20.8%	#3 of 6	Lab model card
Video SimpleQA	71.9%	#2 of 4	Lab model card
Complex Workflow (ByteDance Seed in-house)	64.7%	#1 of 2	Lab model card
LPFQA	52.6%	#1 of 2	Lab model card
MARS-Bench	85.6%	#1 of 2	Lab model card
Inverse IFEval	78.9%	#2 of 3	Lab model card
SimpleQA Verified	36%	#15 of 22	Lab model card
MRCR v2	54%	#21 of 30	Lab model card
MMLongBench	74.8%	#3 of 4	Lab model card
Extended NYT Connections	28.4%	#82 of 103	Official leaderboard
LongFact	92.8%	#4 of 5	Lab model card
LongFact-Concepts	92.8%	#4 of 4	Lab model card
LongFact-Objects	92.9%	#4 of 4	Lab model card
Chinese Complex Instruction Following (ByteDance Seed in-house)	75.3%	#1 of 1	Lab model card
CodeSimpleQA	58%	#1 of 1	Lab model card
Encyclo-K	65.7%	#1 of 1	Lab model card
FActScore precision	71.2%	#1 of 1	Lab model card

Math and science

Benchmark	Score	Rank	Source
AIME 2025	98.3%	#14 of 143	Lab model card
Beyond AIME	86.5%	#2 of 14	Lab model card
HMMT 2025	97.3%	#8 of 54	Lab model card
GPQA Diamond	88.9%	#45 of 302	Lab model card
IMO-AnswerBench	89.3%	#6 of 37	Lab model card
KOR-Bench	77.5%	#1 of 4	Lab model card
IMO 2025	83.3%	#2 of 6	Lab model card
CMO 2025	114.0	#1 of 3	Lab model card
HMMT Nov 2025	93.3%	#10 of 28	Lab model card
PHYBench	74%	#2 of 5	Lab model card
AIME 2026	94.2%	#16 of 38	Lab model card
FrontierScience	74%	#4 of 8	Lab model card
ARC-AGI	85.4%	#11 of 21	Lab model card
Echo General AI Prediction Leaderboard	996	#11 of 17	Official leaderboard
PhyX	72.1%	#2 of 3	Lab model card
ARC-AGI-2	37.5%	#19 of 24	Lab model card
xbench-ScienceQA	69.2%	#23 of 29	Official leaderboard
PutnamBench	35.5%	#6 of 7	Lab model card
MathArena (Apex, IMO, USAMO)	20.3%	#12 of 12	Lab model card
FrontierScience-Research	25%	#5 of 5	Lab model card
HiPhO	74.1%	#4 of 4	Lab model card
FrontierScience-Olympiad	74%	#2 of 2	Lab model card
ProcBench	96.6%	#1 of 1	Lab model card

Coding

Benchmark	Score	Rank	Source
Codeforces rating	3,020	#5 of 47	Lab model card
LiveCodeBench v6	87.8%	#11 of 90	Lab model card
Aider Polyglot	80%	#4 of 30	Lab model card
SWE-bench Verified	76.5%	#37 of 155	Lab model card
SciCode	48.5%	#13 of 49	Lab model card
Multi-SWE-bench	45.2%	#4 of 10	Lab model card
ArtifactsBench	66.6%	#2 of 5	Lab model card
AInstein-SWE-Bench	47.7%	#1 of 2	Lab model card
SWE-bench Multilingual	71.7%	#41 of 73	Lab model card
OJBench	54.9%	#5 of 8	Official leaderboard
AetherCode	60.6%	#2 of 3	Lab model card
SWE-Lancer	49.4%	#8 of 10	Lab model card
Vision2Web S2-2026	21.5%	#8 of 9	Official leaderboard
SWE-Bench Pro	46.9%	#73 of 81	Lab model card
NL2Repo-Bench	27.9%	#29 of 30	Lab model card
EvoCode-Bench	1.90%	#13 of 13	Official leaderboard
HarnessDev	29.8%	#6 of 6	Official leaderboard
SWE-EVO	8.50%	#1 of 1	Lab model card

Agents and tools

Benchmark	Score	Rank	Source
BrowseComp-ZH	82.4%	#1 of 19	Lab model card
BFCL V4	73.4%	#2 of 30	Lab model card
MobileWorld	50.2%	#2 of 18	Official leaderboard
VitaBench 2.0	47.4%	#2 of 17	Official leaderboard
tau2-bench	92.3%	#12 of 79	Lab model card
VITA-Bench	47%	#3 of 16	Lab model card
GameHorizon	71.4	#12 of 37	Official leaderboard
LisanBench	1753.0	#47 of 133	Official leaderboard
ClawProBench	61.1%	#18 of 49	Official leaderboard
MM-BrowseComp	48.8%	#2 of 5	Lab model card
BrowseComp	77.3%	#37 of 81	Lab model card
AutoResearchBench	6.80%	#6 of 13	Official leaderboard
Terminal-Bench 2.0	55.8%	#30 of 63	Lab model card
MobileGym	52%	#2 of 4	Official leaderboard
Customer Support Q&A (ByteDance Seed in-house)	72.4%	#1 of 2	Lab model card
DeepConsult	61.1%	#1 of 2	Lab model card
DeepResearch Bench	53.3%	#1 of 2	Lab model card
MCPMark	54.7%	#6 of 10	Lab model card
DeepSearchQA	77.4%	#16 of 23	Lab model card
DeepPlanning	21.6%	#11 of 15	Official leaderboard
PACT	1,491	#21 of 26	Official leaderboard
TERMS-Bench	0.52	#11 of 11	Official leaderboard
ResearchRubrics	50.7%	#6 of 6	Lab model card
DeR2 Bench	58.2%	#1 of 1	Lab model card
MineDojo	49%	#1 of 1	Lab model card

Professional work

Benchmark	Score	Rank	Source
FinSearchComp	70.2%	#2 of 7	Lab model card
XpertBench	64.5%	#1 of 3	Lab model card
Education (ByteDance Seed in-house)	62.8%	#1 of 2	Lab model card
Information Extraction (ByteDance Seed in-house)	52%	#1 of 2	Lab model card
SpreadsheetBench Verified	79.1%	#1 of 2	Lab model card
HealthBench	57.7%	#20 of 31	Lab model card
FORTE	57.7%	#18 of 21	Official leaderboard
HealthBench Hard	29.1%	#19 of 21	Lab model card
Compositional Tasks (ByteDance Seed in-house)	59.1%	#1 of 1	Lab model card
GDPVal-Diamond	21.3%	#1 of 1	Lab model card
Text Classification (ByteDance Seed in-house)	69%	#1 of 1	Lab model card

Vision, audio and robotics

Benchmark	Score	Rank	Source
MMMU	85.4%	#3 of 103	Lab model card
MeasureBench Real-world	41.7%	#1 of 19	Official leaderboard
MeasureBench Synthetic	39.2%	#1 of 19	Official leaderboard
MathVista	89.8%	#4 of 70	Lab model card
Video-MME	89.5%	#3 of 51	Lab model card
VideoMME v2 (with subtitles)	43.3%	#2 of 31	Official leaderboard
BLINK	79.5%	#2 of 29	Lab model card
VLMs Are Blind	98.6%	#1 of 14	Lab model card
MuirBench	81.8%	#1 of 12	Lab model card
MMStar	83%	#4 of 36	Lab model card
MotionBench	75.2%	#1 of 9	Lab model card
Vibe-Eval	81.4%	#1 of 9	Lab model card
RefSpatial-Bench	72.6%	#2 of 16	Lab model card
Video-MMMU	86.9%	#5 of 36	Lab model card
LogicVista	81.9%	#2 of 14	Lab model card
TempCompass	89.6%	#1 of 7	Lab model card
RealWorldQA	86%	#6 of 38	Lab model card
All-Angles Bench	72.1%	#1 of 6	Lab model card
LongVideoBench	80.3%	#2 of 11	Lab model card
MMVU	78.2%	#5 of 26	Lab model card
EgoTempo	71.8%	#1 of 5	Lab model card
MMLongBench-Doc	61.4%	#4 of 19	Lab model card
MATH-Vision (MathVision)	88.8%	#11 of 51	Lab model card
ERQA	68.5%	#9 of 36	Lab model card
CiteVQA	44.1%	#4 of 16	Official leaderboard
ChartQA Pro	71.2%	#2 of 8	Lab model card
CC-OCR-Bench-V2	73%	#4 of 15	Official leaderboard
CharXiv (DQ)	93.5%	#2 of 7	Lab model card
TreeBench	64.7%	#2 of 7	Lab model card
VisuLogic	47.4%	#2 of 7	Lab model card
MMMU-Pro	78.2%	#26 of 90	Lab model card
LVBench	76.4%	#9 of 31	Lab model card
Vision Arena	1,256	#40 of 136	Official leaderboard
MTVQA-Test	51.1%	#2 of 6	Lab model card
Video-Holmes	67.4%	#2 of 6	Lab model card
ContPhy	67.4%	#1 of 3	Lab model card
Point-Bench	81.4%	#1 of 3	Lab model card
VideoReasonBench	77.8%	#1 of 3	Lab model card
CountBench	95.5%	#11 of 28	Lab model card
OVBench	69.2%	#2 of 5	Lab model card
OVO-Bench	77%	#2 of 5	Lab model card
Scientists' First Exam (SFE)	55.6%	#2 of 5	Lab model card
BabyVision	60.6%	#11 of 27	Official leaderboard
CharXiv	80.5%	#33 of 75	Lab model card
OCRBench v2	62.5%	#10 of 22	Lab model card
MDPBench	74.2	#6 of 12	Official leaderboard
DUDE	72.4%	#2 of 4	Lab model card
LiveSports-3K	78%	#2 of 4	Lab model card
XLRS-Bench	54.6%	#2 of 4	Lab model card
MME-CC	57%	#1 of 2	Lab model card
MORSE-500	37.4%	#1 of 2	Lab model card
ODV-Bench	72.5%	#1 of 2	Lab model card
VPCT	76%	#1 of 2	Lab model card
ZeroVideo	14.5%	#1 of 2	Lab model card
DynaMath	68.9%	#14 of 23	Lab model card
MINERVA	66.5%	#4 of 6	Lab model card
MMSI-Bench	32.5%	#4 of 6	Lab model card
CG-Bench	65%	#2 of 3	Lab model card
EMMA	72%	#2 of 3	Lab model card
HLE-VL	39.2%	#2 of 3	Lab model card
TVBench	75%	#5 of 7	Lab model card
WorldVQA	49.9%	#7 of 9	Lab model card
TOMATO	59.9%	#4 of 5	Lab model card
VLMsAreBiased	77.4%	#4 of 5	Lab model card
ZeroBench	12%	#25 of 31	Lab model card
VisFactor	36.8%	#4 of 4	Lab model card
CrossVid	60.3%	#3 of 3	Lab model card
ViSpeak-Bench	78.5%	#3 of 3	Lab model card
VideoEval-Pro	48%	#2 of 2	Lab model card
LongDocURL	74.7%	#1 of 1	Lab model card
Math Kangaroo 2025 (MathArena)	90.5%	#1 of 1	Lab model card
MathCanvas-Bench	61.9%	#1 of 1	Lab model card
ViVerBench	75.9%	#1 of 1	Lab model card

Safety and security

Benchmark	Score	Rank	Source
BIOBench (ByteDance Seed)	53.5%	#1 of 2	Lab model card
SUPERChem	51.6%	#5 of 5	Lab model card
BABE	50%	#2 of 2	Lab model card

Its lab's numbers put it at 149.9, number 94 of 404

The Self-Reported Index fits one score to the 71 capability results ByteDance Seed reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 148.7 to 151.2. How we compute it.

Its lab's numbers next to 3 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark	Lab says	Board says	Gap
BabyVision	60.6% BabyVision	60.6% Leaderboard, high	Same
MATH-Vision (MathVision)	88.8% MathVision	88.8% Leaderboard	Same
MMLU-Pro	87% MMLU-Pro	87% Leaderboard	Same