跳转到主要内容

Qwen3 235B A22B 2507 Instruct

AlibabaQwen开源权重Apache 2.0 · 商用许可

描述

Qwen3-235B-A22B-Instruct-2507 is the updated instruct version of Qwen3-235B-A22B featuring significant improvements in general capabilities including instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage. It provides substantial gains in long-tail knowledge coverage across multiple languages and markedly better alignment with user preferences in subjective and open-ended tasks.

发布日期
2025-07-21
参数规模
235.0B
上下文长度
262K
支持模态
text

能力雷达图

36
general
49
coding
76
reasoning
49
science估算
60
agents
0
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

领域#排名分数来源
智能体能力模型榜11
71.0
LS
代码能力榜235
41.0
AA
通用能力榜229
46.0
AA
数学推理96
78.0
AA
推理能力80
55.0
LS
科学能力172
52.0
AA

基准测试分数 (LLM Stats)

Agents

BFCL-v370.9%自报

Biology

GPQA77.5%自报

Chemistry

SuperGPQA62.6%自报

Code

Aider-Polyglot57.3%自报

Communication

WritingBench85.2%自报
Multi-IF77.5%自报
Tau2 Retail71.3%自报
Tau2 Airline44.0%自报

Creativity

Creative Writing v387.5%自报
Arena-Hard v279.2%自报

Factuality

SimpleQA54.3%自报

Finance

MMLU-Pro83.0%自报
MMLU-ProX79.4%自报

General

MMLU-Redux93.1%自报
IFEval88.7%自报
MultiPL-E87.9%自报
CSimpleQA84.3%自报
Include79.5%自报
LiveBench 2024112575.4%自报
LiveCodeBench v651.8%自报

Math

AIME 202570.3%自报
HMMT2555.4%自报
PolyMATH50.2%自报

Reasoning

ZebraLogic95.0%自报
ARC-AGI41.8%自报

AA 评测指数

Math Index
71.7
Intelligence Index
18.2
Math 500
1.0
Mmlu Pro
0.8
Gpqa
0.8
Aime
0.7
Aime 25
0.7
Livecodebench
0.5
Ifbench
0.5
Scicode
0.4
Tau2
0.3
Lcr
0.3
Terminalbench Hard
0.2
Hle
0.1

LLM Stats 分类评分

Instruction Following
80
Language
80
Legal
80
Structured Output
80
Finance
80
Healthcare
80
Biology
80
Creativity
80
Writing
80
Math
70
Physics
70
Reasoning
70
General
70
Agents
70
Chemistry
70
Communication
70
Code
60
Economics
60
Tool Calling
60
Multimodal
50
Spatial Reasoning
50
Factuality
50
Vision
50

定价

输入价格$0.2 / 1M tokens
输出价格$0.825 / 1M tokens
混合价格(3:1)$0.356 / 1M tokens

速度

Tokens/秒68.9
首Token延迟1.09s
首回答延迟1.09s

供应商价格排行

供应商价格排行

23 个供应商

最便宜: Cortecs最贵: Scaleway
供应商输入输出
1Cortecs最便宜
$0.062
$0.408
2SiliconFlow (China)
$0.09
$0.6
3NovitaAI
$0.09
$0.58
4Meganova
$0.09
$0.6
5LLM Gateway
$0.09
$0.58
6OpenRouter
$0.1
$0.1
7Weights & Biases
$0.1
$0.1
8IO.NET
$0.11
$0.6
9Chutes
$0.11
$0.6
10Kilo Gateway
$0.11
$0.6
11Abacus
$0.13
$0.6
12SiliconFlow
$0.13
$0.6
13Jiekou.AI
$0.15
$0.8
14Venice AI
$0.15
$0.75
15Alibaba主要
$0.2
$0.825
16submodel
$0.2
$0.6
17Nebius Token Factory
$0.2
$0.6
18Friendli
$0.2
$0.8
19302.AI
$0.29
$1.143
20NanoGPT
$0.3
$0.5
21Hugging Face
$0.3
$3
22Synthetic
$0.65
$3
23Scaleway
$0.75
$2.25

比较该模型在不同 API 供应商之间的定价。

外部链接