Seed 2.1 Pro
ByteDanceProprietary
描述
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.
发布日期
2026-06-24
参数规模
—
上下文长度
—
支持模态
—
能力雷达图
80
general
60
coding
70
reasoning
51
science估算
70
agents
70
multimodal
Science 在缺少专门科学评测时使用推理能力代理估算。
排行榜排名
基准测试分数 (LLM Stats)
3d
BLINK
81.4%自报
Agents
GDPval
87.9%自报
BrowseComp
86.2%自报
MCP Atlas
83.8%自报
OSWorld
78.8%自报
Web Bench
78.4%自报
MobileWorld
73.1%自报
OfficeQA Pro
72.2%自报
Terminal-Bench 2.1
71.0%自报
CyberGym
70.2%自报
OneMillion Bench
68.8%自报
Agent Startup Bench
68.8%自报
SeedClawBench
66.6%自报
Trae Error Fix
63.3%自报
Trae Code Gen
62.4%自报
WildClawBench
61.7%自报
xDailyBench
61.0%自报
Finance Agent v1.1
60.7%自报
SWE-Bench Pro
57.5%自报
Repo Env
55.0%自报
PresentBench
54.6%自报
Workspace Bench
53.0%自报
Doubao Multi-Turn Bench
52.5%自报
ClawEval-MM
51.0%自报
Toolathlon
50.6%自报
Program Bench
50.3%自报
NL2Repo
47.0%自报
CreativeWork
42.5%自报
Agents' Last Exam
41.4%自报
SWE-Atlas
35.2%自报
APEX-Agents
33.8%自报
DeepSWE
32.7%自报
GameWorld
31.2%自报
PostTrainBench
16.5%自报
Biology
SciCode
59.8%自报
Chemistry
SuperGPQA
70.8%自报
SuperChem
59.8%自报
Code
Artifacts Bench
51.0%自报
FrontierCS
46.3%自报
Coding
AetherCode
65.8%自报
Image2FloorPlan
48.0%自报
Embodied
EmbSpatialBench
0.83 / 100自报
General
MMMU-Pro
82.7%自报
SimpleVQA
0.74 / 100自报
MSQA
50.2%自报
KINA
48.3%自报
Image To Text
OCRBench_V2
63.2%自报
Knowledge
VideoSimpleQA
76.4%自报
WorldBench
67.6%自报
Long Context
DUDE
82.8%自报
LongVideoBench
80.6%自报
MMLongBench-128K
78.3%自报
LVBench
78.0%自报
Math
MathVision
94.5%自报
MathVista
90.7%自报
MathVerse
89.7%自报
Beyond AIME
87.0%自报
EMMA
79.3%自报
FrontierScience Olympiad
75.0%自报
DynaMath
73.1%自报
IMO 2025
0.65 / 42自报
Humanity's Last Exam
55.7%自报
IMOProof-Adv
54.3%自报
MathArena Apex
31.3%自报
LiveMathematicianBench
20.9%自报
HorizonMath
2.0%自报
Multimodal
CharXiv-D
95.5%自报
Video-MME
89.2%自报
CharXiv-R
86.4%自报
VLMsAreBiased
83.6%自报
OVOBench
80.7%自报
TVBench
80.5%自报
TOMATO
79.5%自报
LiveSports-3K
76.8%自报
MotionBench
74.9%自报
BabyVision
73.7%自报
TreeBench
71.1%自报
ChartQAPro
70.9%自报
Minerva
70.7%自报
OVBench
70.0%自报
VideoHolmes
68.2%自报
CrossVid
65.0%自报
ContPhy
63.6%自报
MeasureBench
62.9%自报
ZEROBench
0.56 / 100自报
VisuLogic
0.54 / 100自报
WorldVQA
53.0%自报
VisFactor
51.4%自报
MMSIBench
35.9%自报
Physics
IPhO 2025
79.3%自报
Reasoning
ERQA
72.0%自报
ArcAGI2
62.5%自报
FrontierScience Research
28.3%自报
Spatial Reasoning
RealWorldQA
86.7%自报
AA 评测指数
暂无 AA 评测数据
LLM Stats 分类评分
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
定价
暂无定价数据
速度
暂无速度数据
供应商价格排行
暂无提供商数据