Anthropic

由前 OpenAI 核心团队创立的 AI 安全公司，总部旧金山。Claude 系列以卓越的指令遵循、超长文本处理和代码生成能力著称，是开发者最信赖的编程助手之一。

Models

所有模型

模型名称	上下文	智能指数
Claude 2.0	—	9.1
Claude 2.1	—	9.3
Claude 3 Haiku	200K	12.3
Claude 3 Opus	—	18.0
Claude 3 Sonnet	200K	10.3
Claude 3.5 Haiku	200K	18.7
Claude 3.5 Sonnet (June '24)	—	14.2
Claude 3.5 Sonnet (Oct '24)	200K	15.9
Claude 3.7 Sonnet (Non-reasoning)	200K	30.8
Claude 3.7 Sonnet (Reasoning)	200K	34.7
Claude 4 Opus (Non-reasoning)	1M	33.0
Claude 4 Opus (Reasoning)	—	39.0
Claude 4 Sonnet (Non-reasoning)	1M	33.0
Claude 4 Sonnet (Reasoning)	—	38.7
Claude 4.1 Opus (Non-reasoning)	200K	36.0
Claude 4.1 Opus (Reasoning)	—	42.0
Claude 4.5 Haiku (Non-reasoning)	200K	31.1
Claude 4.5 Haiku (Reasoning)	—	37.1
Claude 4.5 Sonnet (Non-reasoning)	1M	37.1
Claude 4.5 Sonnet (Reasoning)	—	43.0
Claude Instant	—	7.4
Claude Opus 4.5 (Non-reasoning)	200K	43.1
Claude Opus 4.5 (Reasoning)	200K	49.7
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)	1M	53.0
Claude Opus 4.6 (Non-reasoning, High Effort)	1M	46.5
Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)	1M	51.7
Claude Sonnet 4.6 (Non-reasoning, High Effort)	1M	44.4
Claude Sonnet 4.6 (Non-reasoning, Low Effort)	1M	42.6