Seed 2.1 Turbo

ByteDanceProprietary

描述

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.

发布日期

2026-06-24

参数规模

—

上下文长度

262K

支持模态

image, text, video

能力雷达图

general

coding

reasoning

science估算

agents

multimodal

缺少专门科学评测时，Science 由 LLM Stats 科学得分或推理能力估算。

排行榜排名

领域	#排名	分数	来源
智能体能力模型榜	61	50.0	LS
多模态榜	25	58.0	LS

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

3d

BLINK

79.4%自报

Agents

BrowseCompOpenAI (2025)

84.9%自报

GDPval

82.7%自报

MCP Atlas

80.3%自报

OSWorld

76.4%自报

Web Bench

73.6%自报

OfficeQA Pro

71.1%自报

MobileWorld

70.0%自报

Terminal-Bench 2.1

67.6%自报

CyberGym

67.0%自报

OneMillion Bench

66.6%自报

SeedClawBench

63.8%自报

WildClawBench

62.8%自报

Trae Code Gen

59.7%自报

SWE-Bench ProPrinceton NLP (2024)

57.0%自报

Trae Error Fix

56.7%自报

xDailyBench

56.4%自报

Finance Agent v1.1

56.0%自报

Workspace Bench

54.7%自报

Agent Startup Bench

54.0%自报

Program Bench

49.4%自报

Toolathlon

49.1%自报

Doubao Multi-Turn Bench

49.0%自报

PresentBench

48.3%自报

Repo Env

46.7%自报

ClawEval-MM

46.0%自报

NL2Repo

43.7%自报

CreativeWork

34.5%自报

SWE-Atlas

30.6%自报

APEX-Agents

29.2%自报

GameWorld

25.9%自报

DeepSWE

23.0%自报

PostTrainBench

18.3%自报

Biology

SciCode

57.8%自报

Chemistry

SuperGPQA

67.4%自报

SuperChem

56.6%自报

Code

FrontierCS

50.8%自报

Artifacts Bench

47.0%自报

Image2FloorPlan

35.9%自报

Coding

AetherCode

67.9%自报

Embodied

EmbSpatialBench

0.82 / 100自报

General

MMMU-Pro

82.2%自报

SimpleVQA

0.71 / 100自报

KINA

46.6%自报

MSQA

42.0%自报

Image To Text

OCRBench_V2

62.8%自报

Knowledge

VideoSimpleQA

71.4%自报

WorldBench

63.7%自报

Long Context

DUDE

83.1%自报

LongVideoBench

80.6%自报

MMLongBench-128K

76.9%自报

LVBench

76.8%自报

Math

MathVision

92.7%自报

MathVista

90.5%自报

MathVerse

89.2%自报

Beyond AIME

88.0%自报

EMMA

78.4%自报

FrontierScience Olympiad

76.0%自报

DynaMath

68.1%自报

Humanity's Last Exam

54.6%自报

MathArena Apex

35.4%自报

LiveMathematicianBench

27.7%自报

HorizonMath

2.0%自报

Multimodal

CharXiv-D

94.6%自报

Video-MME

89.0%自报

CharXiv-R

83.6%自报

OVOBench

79.2%自报

TVBench

77.2%自报

LiveSports-3K

77.1%自报

MotionBench

74.8%自报

TreeBench

71.1%自报

ChartQAPro

70.9%自报

OVBench

69.7%自报

VLMsAreBiased

68.3%自报

VideoHolmes

67.6%自报

Minerva

65.9%自报

CrossVid

63.2%自报

BabyVision

62.9%自报

ContPhy

61.1%自报

MeasureBench

58.9%自报

TOMATO

56.8%自报

VisuLogic

0.53 / 100自报

WorldVQA

48.6%自报

VisFactor

43.9%自报

MMSIBench

31.4%自报

ZEROBench

0.11 / 100自报

Reasoning

ERQA

71.3%自报

ArcAGI2

61.3%自报

FrontierScience Research

33.3%自报

Spatial Reasoning

RealWorldQA

86.3%自报

AA 评测指数

(Artificial Analysis)

暂无 AA 评测数据

LLM Stats 分类评分

(LLM Stats (zeroeval))

Structured Output

Legal

Spatial Reasoning

Embodied

Finance

Long Context

Multimodal

Safety

Image To Text

Healthcare

Economics

Tool Calling

Video

Vision

Math

Physics

Reasoning

Knowledge

General

Agents

Biology

Chemistry

Coding

Science

Frontend Development

Code

Systems

定价

输入价格$0.5 / 1M tokens

输出价格$2.5 / 1M tokens

混合价格(3:1)$1 / 1M tokens

速度

暂无速度数据

供应商价格排行

4 个供应商

最便宜: ByteDance最贵: Kilo Gateway

供应商输入输出

1ByteDance主要

$0.5

$2.5

2NanoGPT

$0.5

$2.5

3OpenRouter

$0.5

$2.5

4Kilo Gateway

$0.5

$2.5

比较该模型在不同 API 供应商之间的定价。

外部链接

LLM Stats Artificial Analysis