Seed 2.1 Turbo

ByteDanceProprietary

Description

ByteDance's low-cost, low-latency next-generation agent model for large-scale production. A deep-thinking model with the full feature set of Seed 2.1 Pro and comparable performance, it is designed for enterprise-grade deployments handling high volumes of online calls across coding, long-chain agents, and multimodal tasks. Served via Volcano Engine as Doubao-Seed-2.1-turbo.

Date de sortie

2026-06-24

Paramètres

—

Longueur du contexte

—

Modalités

—

Radar de capacités

general

coding

reasoning

scienceest.

agents

multimodal

Science utilise un proxy de raisonnement lorsque les benchmarks scientifiques dédiés ne sont pas disponibles.

Classements

Domaine	#Rang	Score	Source
Capacité agentique	51	57.0	LS
Classement multimodal	76	66.0	LS
Raisonnement	73	57.0	LS

Scores de benchmarks (LLM Stats)

3d

BLINK

79.4%Aut.

Agents

BrowseComp

84.9%Aut.

GDPval

82.7%Aut.

MCP Atlas

80.3%Aut.

OSWorld

76.4%Aut.

Web Bench

73.6%Aut.

OfficeQA Pro

71.1%Aut.

MobileWorld

70.0%Aut.

Terminal-Bench 2.1

67.6%Aut.

CyberGym

67.0%Aut.

OneMillion Bench

66.6%Aut.

SeedClawBench

63.8%Aut.

WildClawBench

62.8%Aut.

Trae Code Gen

59.7%Aut.

SWE-Bench Pro

57.0%Aut.

Trae Error Fix

56.7%Aut.

xDailyBench

56.4%Aut.

Finance Agent v1.1

56.0%Aut.

Workspace Bench

54.7%Aut.

Agent Startup Bench

54.0%Aut.

Program Bench

49.4%Aut.

Toolathlon

49.1%Aut.

Doubao Multi-Turn Bench

49.0%Aut.

PresentBench

48.3%Aut.

Repo Env

46.7%Aut.

ClawEval-MM

46.0%Aut.

NL2Repo

43.7%Aut.

CreativeWork

34.5%Aut.

SWE-Atlas

30.6%Aut.

APEX-Agents

29.2%Aut.

GameWorld

25.9%Aut.

DeepSWE

23.0%Aut.

PostTrainBench

18.3%Aut.

Biology

SciCode

57.8%Aut.

Chemistry

SuperGPQA

67.4%Aut.

SuperChem

56.6%Aut.

Code

FrontierCS

50.8%Aut.

Artifacts Bench

47.0%Aut.

Coding

AetherCode

67.9%Aut.

Image2FloorPlan

35.9%Aut.

Embodied

EmbSpatialBench

0.82 / 100Aut.

General

MMMU-Pro

82.2%Aut.

SimpleVQA

0.71 / 100Aut.

KINA

46.6%Aut.

MSQA

42.0%Aut.

Image To Text

OCRBench_V2

62.8%Aut.

Knowledge

VideoSimpleQA

71.4%Aut.

WorldBench

63.7%Aut.

Long Context

DUDE

83.1%Aut.

LongVideoBench

80.6%Aut.

MMLongBench-128K

76.9%Aut.

LVBench

76.8%Aut.

Math

MathVision

92.7%Aut.

MathVista

90.5%Aut.

MathVerse

89.2%Aut.

Beyond AIME

88.0%Aut.

EMMA

78.4%Aut.

FrontierScience Olympiad

76.0%Aut.

DynaMath

68.1%Aut.

Humanity's Last Exam

54.6%Aut.

MathArena Apex

35.4%Aut.

LiveMathematicianBench

27.7%Aut.

HorizonMath

2.0%Aut.

Multimodal

CharXiv-D

94.6%Aut.

Video-MME

89.0%Aut.

CharXiv-R

83.6%Aut.

OVOBench

79.2%Aut.

TVBench

77.2%Aut.

LiveSports-3K

77.1%Aut.

MotionBench

74.8%Aut.

TreeBench

71.1%Aut.

ChartQAPro

70.9%Aut.

OVBench

69.7%Aut.

VLMsAreBiased

68.3%Aut.

VideoHolmes

67.6%Aut.

Minerva

65.9%Aut.

CrossVid

63.2%Aut.

BabyVision

62.9%Aut.

ContPhy

61.1%Aut.

MeasureBench

58.9%Aut.

ZEROBench

0.57 / 100Aut.

TOMATO

56.8%Aut.

VisuLogic

0.53 / 100Aut.

WorldVQA

48.6%Aut.

VisFactor

43.9%Aut.

MMSIBench

31.4%Aut.

Reasoning

ERQA

71.3%Aut.

ArcAGI2

61.3%Aut.

FrontierScience Research

33.3%Aut.

Spatial Reasoning

RealWorldQA

86.3%Aut.

Indices d'évaluation AA

Aucune donnée d'évaluation AA disponible

Scores par catégorie LLM Stats

Structured Output

Legal

Long Context

Spatial Reasoning

Embodied

Finance

General

Image To Text

Math

Multimodal

Safety

Healthcare

Economics

Tool Calling

Video

Vision

Physics

Reasoning

Agents

Biology

Chemistry

Frontend Development

Code

Coding

Science

Systems

Tarification

Aucune donnée de prix disponible

Vitesse

Aucune donnée de vitesse disponible

Classement des Prix par Fournisseur

Aucune donnée de fournisseur disponible

Sources externes

LLM Stats Artificial Analysis