Seed 2.1 Pro
ByteDanceProprietary
Описание
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.
Дата выхода
2026-06-24
Параметры
—
Длина контекста
—
Модальности
—
Радар способностей
80
general
60
coding
70
reasoning
51
scienceоцен.
70
agents
70
multimodal
Science использует прокси на основе рассуждений, когда специализированные научные бенчмарки недоступны.
Рейтинги
| Домен | #Место | Оценка | Источник |
|---|---|---|---|
| Агентные возможности | 38 | 60.0 | LS |
| Мультимодальный рейтинг | 70 | 70.0 | LS |
| Рассуждения | 79 | 56.0 | LS |
Оценки бенчмарков (LLM Stats)
3d
BLINK
81.4%Сам.
Agents
GDPval
87.9%Сам.
BrowseComp
86.2%Сам.
MCP Atlas
83.8%Сам.
OSWorld
78.8%Сам.
Web Bench
78.4%Сам.
MobileWorld
73.1%Сам.
OfficeQA Pro
72.2%Сам.
Terminal-Bench 2.1
71.0%Сам.
CyberGym
70.2%Сам.
OneMillion Bench
68.8%Сам.
Agent Startup Bench
68.8%Сам.
SeedClawBench
66.6%Сам.
Trae Error Fix
63.3%Сам.
Trae Code Gen
62.4%Сам.
WildClawBench
61.7%Сам.
xDailyBench
61.0%Сам.
Finance Agent v1.1
60.7%Сам.
SWE-Bench Pro
57.5%Сам.
Repo Env
55.0%Сам.
PresentBench
54.6%Сам.
Workspace Bench
53.0%Сам.
Doubao Multi-Turn Bench
52.5%Сам.
ClawEval-MM
51.0%Сам.
Toolathlon
50.6%Сам.
Program Bench
50.3%Сам.
NL2Repo
47.0%Сам.
CreativeWork
42.5%Сам.
Agents' Last Exam
41.4%Сам.
SWE-Atlas
35.2%Сам.
APEX-Agents
33.8%Сам.
DeepSWE
32.7%Сам.
GameWorld
31.2%Сам.
PostTrainBench
16.5%Сам.
Biology
SciCode
59.8%Сам.
Chemistry
SuperGPQA
70.8%Сам.
SuperChem
59.8%Сам.
Code
Artifacts Bench
51.0%Сам.
FrontierCS
46.3%Сам.
Coding
AetherCode
65.8%Сам.
Image2FloorPlan
48.0%Сам.
Embodied
EmbSpatialBench
0.83 / 100Сам.
General
MMMU-Pro
82.7%Сам.
SimpleVQA
0.74 / 100Сам.
MSQA
50.2%Сам.
KINA
48.3%Сам.
Image To Text
OCRBench_V2
63.2%Сам.
Knowledge
VideoSimpleQA
76.4%Сам.
WorldBench
67.6%Сам.
Long Context
DUDE
82.8%Сам.
LongVideoBench
80.6%Сам.
MMLongBench-128K
78.3%Сам.
LVBench
78.0%Сам.
Math
MathVision
94.5%Сам.
MathVista
90.7%Сам.
MathVerse
89.7%Сам.
Beyond AIME
87.0%Сам.
EMMA
79.3%Сам.
FrontierScience Olympiad
75.0%Сам.
DynaMath
73.1%Сам.
IMO 2025
0.65 / 42Сам.
Humanity's Last Exam
55.7%Сам.
IMOProof-Adv
54.3%Сам.
MathArena Apex
31.3%Сам.
LiveMathematicianBench
20.9%Сам.
HorizonMath
2.0%Сам.
Multimodal
CharXiv-D
95.5%Сам.
Video-MME
89.2%Сам.
CharXiv-R
86.4%Сам.
VLMsAreBiased
83.6%Сам.
OVOBench
80.7%Сам.
TVBench
80.5%Сам.
TOMATO
79.5%Сам.
LiveSports-3K
76.8%Сам.
MotionBench
74.9%Сам.
BabyVision
73.7%Сам.
TreeBench
71.1%Сам.
ChartQAPro
70.9%Сам.
Minerva
70.7%Сам.
OVBench
70.0%Сам.
VideoHolmes
68.2%Сам.
CrossVid
65.0%Сам.
ContPhy
63.6%Сам.
MeasureBench
62.9%Сам.
ZEROBench
0.56 / 100Сам.
VisuLogic
0.54 / 100Сам.
WorldVQA
53.0%Сам.
VisFactor
51.4%Сам.
MMSIBench
35.9%Сам.
Physics
IPhO 2025
79.3%Сам.
Reasoning
ERQA
72.0%Сам.
ArcAGI2
62.5%Сам.
FrontierScience Research
28.3%Сам.
Spatial Reasoning
RealWorldQA
86.7%Сам.
Индексы оценки AA
Нет данных AA оценки
Оценки категорий LLM Stats
Structured Output100
Search90
Legal80
Long Context80
Spatial Reasoning80
Embodied80
Finance80
General80
3d80
Image To Text70
Math70
Multimodal70
Physics70
Reasoning70
Safety70
Healthcare70
Chemistry70
Economics70
Tool Calling70
Video70
Vision70
Agents60
Biology60
Code60
Frontend Development50
Coding50
Science30
Systems20
Цены
Нет данных о ценах
Скорость
Нет данных о скорости
Рейтинг цен провайдеров
Нет данных провайдеров