跳转到主要内容

Claude 4.5 Haiku (Non-reasoning)

AnthropicClaudeProprietary

描述

Claude Haiku 4.5 is Anthropic's fastest, most cost-efficient model, matching Sonnet 4's performance on coding, computer use, and agent tasks. It offers similar performance to Sonnet 4 at one-third the cost and more than twice the speed, making it ideal for high-volume, latency-sensitive applications and multi-agent orchestration.

发布日期
2025-10-15
参数规模
上下文长度
200K
支持模态
image, text

能力雷达图

42
general
38
coding
43
reasoning
42
science估算
80
agents
80
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
智能体与工具79
47.0
LS
代码能力榜133
52.0
AA
通用能力榜180
52.0
AA
数学推理223
39.0
AA
科学能力242
43.0
AA

基准测试分数 (LLM Stats)

Agents

OSWorld50.7%自报
Terminal-Bench41.0%自报

Biology

GPQA73.0%自报

Code

SWE-Bench Verified73.3%自报

Communication

Tau2 Retail83.2%自报
Tau2 Telecom83.0%自报
Tau2 Airline63.6%自报

General

MMMLU83.0%自报
MMMU (validation)73.2%自报

Math

AIME 202580.7%自报

Safety

Cybersecurity CTFs46.9%自报

AA 评测指数

Math Index
39.0
Intelligence Index
31.1
Coding Index
29.6
Mmlu Pro
0.8
Gpqa
0.6
Livecodebench
0.5
Lcr
0.4
Ifbench
0.4
Aime 25
0.4
Scicode
0.3
Tau2
0.3
Terminalbench Hard
0.3
Hle
0.0

LLM Stats 分类评分

Tool Calling
80
Communication
80
Language
80
Math
80
Biology
70
Chemistry
70
Frontend Development
70
General
70
Physics
70
Reasoning
70
Vision
60
Code
60
Multimodal
60
Agents
50
Healthcare
50
Safety
50
Long Context
20

定价

输入价格$1.25 / 1M tokens
输出价格$5 / 1M tokens
混合价格(3:1)$2.188 / 1M tokens

速度

Tokens/秒121.7 tokens/s
首Token延迟0.52s
首回答延迟0.52s

可用提供商

(LS 内部计价单位)
提供商输入价格输出价格
Anthropic1.0M5.0M

外部链接