Claude Haiku 5.5
Anthropic · 2026-10-07 · 通用模型
官方发布来源 · 按需求选型 · 查看覆盖缺口 · 归档与转录记录
Claude Haiku 5.5
Anthropic 迄今速度最快、成本最低的小模型,面向高吞吐量与成本敏感的自动化、快速摘要、查询和编码子智能体协作;收录 8 项主干基准,包括 OSWorld 2.1 72.4%、HLE with tools 57.4%、GDPval-AA 1620。
- 输入模态
- 官方资料未说明
- 上下文
- 官方资料未说明
- 参数
- 官方资料未说明
- 价格(每百万 tokens)
- USD 输入 0.1 / 输出 0.5 · prompts <=100k: 输入 $0.10, 输出 $0.50, 缓存读 $0.01; >100k: 输入 $0.50, 输出 $2.50, 缓存读 $0.05
本变体的评测证据
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Knowledge work GDPval-AA v2.1 · quote_snippet: GDPval-AA v2.1 1620
{
"harness": "Artificial Analysis",
"tools": null,
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 735, GPT-6 Luna 1437, Sonnet 5.5 1840(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Knowledge work AA-Briefcase v1.1 · quote_snippet: AA-Briefcase v1.1 1578
{
"harness": "Artificial Analysis",
"tools": null,
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 614, GPT-6 Luna 1336, Sonnet 5.5 1824(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Computer use OSWorld 2.1 · quote_snippet: OSWorld 2.1 72.4% Offline subset
{
"harness": "OSWorld Offline subset",
"tools": null,
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 15.7%, GPT-6 Luna 48.9%, Sonnet 5.5 83.9%(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Multidisciplinary reasoning Humanity’s Last Exam (no tools) · quote_snippet: 45.9% no tools
{
"harness": null,
"tools": [],
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 10.2%, Sonnet 5.5 56.9%(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Multidisciplinary reasoning Humanity’s Last Exam (with tools) · quote_snippet: 57.4% with tools
{
"harness": null,
"tools": [
"search",
"python"
],
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 18.7%, Sonnet 5.5 64.5%(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Agentic coding Terminal-Bench 4.0 · quote_snippet: Terminal-Bench 4.0 39.2%
{
"harness": "Terminal-Bench 4.0",
"tools": null,
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 0.0%, GPT-6 Luna 16.4%, Sonnet 5.5 70.6%(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Agentic coding FrontierCode 1.1 (Main) · quote_snippet: FrontierCode 1.1 (Main) 46.4%
{
"harness": "FrontierCode 1.1",
"tools": null,
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}GPT-6 Luna 42.4%, Sonnet 5.5 52.1% Xhigh(仅参考)
归一化读数:不计算;样本及方差不完整,无法计算置信区间。
原文位置与完整协议
heading: Performance · table: Benchmark comparison grid · row: Visual reasoning Chartography · quote_snippet: Chartography 46.4% no tools
{
"harness": "Surge AI Chartography",
"tools": [],
"shots": null,
"reasoning_effort": null,
"temperature": null,
"top_p": null,
"token_budget": null,
"turn_limit": null,
"time_limit": null,
"run_count": null,
"aggregation": null,
"judge": null
}Haiku 4.5 6.4%, GPT-6 Luna 29.1%, Sonnet 5.5 61.6%(仅参考)