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	—	3.6
Claude 2.1	—	3.9
Claude 3 Haiku	200K	3.9
Claude 3 Opus	—	11.8
Claude 3 Sonnet	200K	4.7
Claude 3.5 Haiku	200K	12.3
Claude 3.5 Sonnet (June '24)	—	8.3
Claude 3.5 Sonnet (Oct '24)	200K	9.9
Claude 3.7 Sonnet (Non-reasoning)	200K	23.5
Claude 3.7 Sonnet (Reasoning)	200K	27.1
Claude 4 Opus (Non-reasoning)	1M	25.5
Claude 4 Opus (Reasoning)	—	31.0
Claude 4 Sonnet (Non-reasoning)	1M	25.5
Claude 4 Sonnet (Reasoning)	—	30.7
Claude 4.1 Opus (Non-reasoning)	200K	28.2
Claude 4.1 Opus (Reasoning)	—	33.7
Claude 4.5 Haiku (Non-reasoning)	200K	23.7
Claude 4.5 Haiku (Reasoning)	—	29.6
Claude 4.5 Sonnet (Non-reasoning)	1M	29.3
Claude 4.5 Sonnet (Reasoning)	—	34.7
Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)	1M	59.9
Claude Instant	—	2.1
Claude Opus 4.5 (Non-reasoning)	200K	34.7
Claude Opus 4.5 (Reasoning)	200K	40.8
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)	1M	43.7
Claude Opus 4.6 (Non-reasoning, High Effort)	1M	37.8
Claude Opus 4.7 (Adaptive Reasoning, Max Effort)	1M	53.5
Claude Opus 4.7 (Non-reasoning, High Effort)	1M	42.7
Claude Opus 4.8 (Adaptive Reasoning, Max Effort)	1M	55.7
Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)	1M	47.2
Claude Sonnet 4.6 (Non-reasoning, High Effort)	1M	35.9
Claude Sonnet 4.6 (Non-reasoning, Low Effort)	1M	34.3