能力前沿
独立追踪 586 款语言模型与 11 个媒体竞技场的能力、价格与速度,每日快照,记录变化。
57.6Anthropic · 2026年9月22日
数据表格 · 25
| 发布 | 模型 | 厂商 | 智能 |
|---|---|---|---|
| 2022-11-30 | GPT-3.5 Turbo | OpenAI | 5.5 |
| 2023-03-14 | GPT-4 | OpenAI | 6.7 |
| 2023-11-06 | GPT-4 Turbo | OpenAI | 7 |
| 2024-03-04 | Claude 3 Opus | Anthropic | 8.7 |
| 2024-09-12 | o1-preview | OpenAI | 11.4 |
| 2024-12-05 | o1 | OpenAI | 15.2 |
| 2025-02-24 | Claude 3.7 Sonnet (Reasoning) | Anthropic | 17.7 |
| 2025-04-16 | o3 | OpenAI | 20.2 |
| 2025-05-22 | Claude 4 Opus (Reasoning) | Anthropic | 20.6 |
| 2025-06-10 | o3-pro | OpenAI | 21.9 |
| 2025-07-10 | Grok 4 | SpaceXAI | 22.5 |
| 2025-08-05 | Claude 4.1 Opus (Reasoning) | Anthropic | 22.8 |
| 2025-08-07 | GPT-5 (high) | OpenAI | 23 |
| 2025-09-23 | GPT-5 Codex (high) | OpenAI | 24.9 |
| 2025-11-18 | Gemini 3 Pro Preview (high) | 28 | |
| 2025-11-24 | Claude Opus 4.5 (Reasoning) | Anthropic | 29.1 |
| 2025-12-11 | GPT-5.2 (xhigh) | OpenAI | 30.4 |
| 2026-02-05 | GPT-5.3 Codex (xhigh) | OpenAI | 32.5 |
| 2026-03-05 | GPT-5.4 (xhigh) | OpenAI | 39 |
| 2026-04-16 | Claude Opus 4.7 (Adaptive Reasoning, Max Effort) | Anthropic | 40.7 |
| 2026-05-28 | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | Anthropic | 41.8 |
| 2026-06-09 | Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Anthropic | 49.6 |
| 2026-07-24 | Claude Opus 5 (Adaptive Reasoning, Max Effort) | Anthropic | 50.8 |
| 2026-09-01 | Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) | Anthropic | 53.4 |
| 2026-09-22 | Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback) | Anthropic | 57.6 |
近期变化
全部变化 →09-26
2026年9月26日1 条变化- 分数Apodex 1.1Apodex智能指数30.4→26.4-13%
09-25
2026年9月25日9 条变化- 新增DeepSeek V4.1 Flash (Non-Reasoning)DeepSeek智能指数 24.7
- 新增GLM-5.3 (low)Z AI智能指数 34.3
- 下架Agnes 2.5 Pro Alpha (based on Qwen3.5 397B A17B)Sapiens AI
- 下架Agnes 2.5 Pro BetaSapiens AI
- 下架Agnes 3.0 FlashSapiens AI
- 价格Gemma 4 26B A4B (Reasoning)Google输入价$0.120→$0.100-17%
- 价格Muse Glimmer (high)Meta输入价$0.350→$0.320-8.6%
- 价格Muse Glimmer (high)Meta输出价$1.50→$1.35-10.0%
- 分数Apodex 1.1Apodex智能体指数首次评测→28.4
09-24
2026年9月24日36 条变化- 新增Mercury 2.5Inception智能指数 12.3
- 价格Devstral 2Mistral输入价$0→指标撤下
- 价格Devstral 2Mistral输出价$0→指标撤下
- 价格Devstral Small 2Mistral输入价$0→指标撤下
- 价格Devstral Small 2Mistral输出价$0→指标撤下
- 价格LFM2.5-8B-A1BLiquid AI输入价$0→指标撤下
- 价格LFM2.5-8B-A1BLiquid AI输出价$0→指标撤下
- 价格LFM2.5-VL-1.6BLiquid AI输入价$0→指标撤下
- 价格LFM2.5-VL-1.6BLiquid AI输出价$0→指标撤下
- 价格Ling-3.0-flash-FinInclusionAI输入价$0→$0.070—
- 价格Ling-3.0-flash-FinInclusionAI输出价$0→$0.220—
- 价格Llama 3.1 Nemotron Instruct 70BNVIDIA输入价$1.20→指标撤下
- 价格Llama 3.1 Nemotron Instruct 70BNVIDIA输出价$1.20→指标撤下
- 价格Llama Nemotron Super 49B v1.5 (Non-reasoning)NVIDIA输入价$0.400→指标撤下
- 价格Llama Nemotron Super 49B v1.5 (Non-reasoning)NVIDIA输出价$0.400→指标撤下
- 价格Llama Nemotron Super 49B v1.5 (Reasoning)NVIDIA输入价$0.400→指标撤下
- 价格Llama Nemotron Super 49B v1.5 (Reasoning)NVIDIA输出价$0.400→指标撤下
- 价格Mistral 7B InstructMistral输入价$0.250→$0.150-40%
- 价格Mistral 7B InstructMistral输出价$0.250→$0.200-20%
- 价格Mistral MediumMistral输入价$1.50→指标撤下
- 价格Mistral MediumMistral输出价$7.50→指标撤下
- 价格Mistral Medium 3.1Mistral输入价$0.400→指标撤下
- 价格Mistral Medium 3.1Mistral输出价$2.00→指标撤下
- 价格Mistral Small (Feb '24)Mistral输入价$0.150→指标撤下
- 价格Mistral Small (Feb '24)Mistral输出价$0.600→指标撤下
- 价格Mistral Small (Sep '24)Mistral输入价$0.200→指标撤下
- 价格Mistral Small (Sep '24)Mistral输出价$0.600→指标撤下
- 价格Mistral Small 3Mistral输入价$0.100→$0.050-50%
- 价格Mistral Small 3Mistral输出价$0.300→$0.080-73%
- 价格Mistral Small 3.1Mistral输入价$0.100→$0.110+10.0%
- 价格Mistral Small 3.1Mistral输出价$0.300→$0.170-43%
- 价格Mistral Small 3.2Mistral输入价$0.100→$0.070-30%
- 价格Mistral Small 3.2Mistral输出价$0.300→$0.200-33%
- 价格NVIDIA Nemotron Nano 12B v2 VL (Reasoning)NVIDIA输入价$0.200→指标撤下
- 价格NVIDIA Nemotron Nano 12B v2 VL (Reasoning)NVIDIA输出价$0.600→指标撤下
- 价格Trinity Large ThinkingArcee AI输出价$0.900→$0.800-11%
- 1Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)Anthropic57.6
- 2Claude Opus 5.5 (Adaptive Reasoning, Xhigh Effort, Default Fallback)Anthropic56
- 3Claude Opus 5.5 (Adaptive Reasoning, High Effort, Default Fallback)Anthropic53.6
- 4Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)Anthropic53.4
- 5Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)Anthropic53.2
- 6GPT-6 Astra (max)OpenAI52.7
- 7GPT-6 Astra (xhigh)OpenAI52.4
- 8Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)Anthropic51.2
- 9Claude Opus 5.5 (Adaptive Reasoning, Medium Effort, Default Fallback)Anthropic51.2
- 10GPT-6 Astra (high)OpenAI50.9
- 文生图GPT Image 2.5 Sunburst (max)162
- 图像编辑GPT Image 2.5 Sunburst (max)77
- 文生视频Wan 3.081
- 图生视频Gemini Omni Flash76
- 文生视频(含声)Gemini Omni Flash36
- 图生视频(含声)Minimax H3 Max (post-trained by fal)35
- 文本转语音Sonic 3.691
- 语音对话StepAudio 3 Realtime44
- 语音转文本GPT-4o Mini Transcribe, OpenAI71
- 纯音乐Mureka V920
- 带人声音乐Suno V5.517
图像、视频、语音与音乐,按 Elo 排名