模型 · 排行榜实时 · 1,269 个模型 · 279 家服务商 · 160 项基准测试
所有 AI 模型 · 已追踪
你真正能用上的最完整 AI 模型清单 · 1,269 个模型, 279 家提供商, 160 项基准测试 · 全部在一处评分、定价和排名。
基准测试评分跨提供商定价上下文 · 参数 · 许可每日更新
并排对比显示 1,269 共 1,269
综合前 10 名
按基准测试平均分排名 · 至少 3 项基准测试
最新动态
新发布、覆盖度、价格领先者和 ELO 冠军
| # | 模型 | 发布时间 |
|---|---|---|
| 1 | H Ampersand v0.1 | Oct 20, 2026 |
| 2 | O Ornith 1.5 122B | Oct 20, 2026 |
| 3 | Oct 19, 2026 | |
| 4 | Oct 19, 2026 | |
| 5 | Oct 19, 2026 | |
| 6 | Oct 17, 2026 | |
| 7 | E Glimpse | Oct 16, 2026 |
| 8 | R Beam 1.0 | Oct 16, 2026 |
| 9 | M Jais 3 0.9B | Oct 15, 2026 |
| 10 | M Jais 3 3.7B | Oct 15, 2026 |
所有模型
| # | 模型 | 供应商 | 类别 | 上下文 | 输入 $/M | 输出 $/M | 平均 | 基准 | ELO | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Anthropic | LLM | 1.0M | n/a | n/a | 98.9 | 14 | n/a | ||
| 2 | OpenAI | LLM | 400K | $30.00 | $180.00 | 98.3 | 3 | n/a | ||
| 3 | OpenAI | 多模态 | 1.1M | $10.00 | $50.00 | 95.8 | 34 | n/a | ||
| 4 | DeepSeek | LLM | 164K | $0.40 | $1.20 | 95.2 | 9 | n/a | ||
| 5 | OpenAI | 多模态 | 1.1M | $2.00 | $10.00 | 93.4 | 25 | n/a | ||
| 6 | Anthropic | 多模态 | 1.0M | $4.00 | $20.00 | 91.8 | 28 | n/a | ||
| 7 | stepfun | LLM | 262K | $0.10 | $0.30 | 89.5 | 10 | n/a | ||
| 8 | Google DeepMind | 多模态 | 1.0M | $2.00 | $10.00 | 89.2 | 24 | n/a | ||
| 9 | OpenAI | 多模态 | 128K | $1.25 | $10.00 | 89.0 | 7 | n/a | ||
| 10 | HA Qwen2.5 72B Instruct Abliterated | HuiHui AI | LLM | n/a | n/a | n/a | 87.2 | 6 | n/a | |
| 11 | DeepSeek | LLM | n/a | n/a | n/a | 85.9 | 7 | n/a | ||
| 12 | OpenAI | 多模态 | 400K | $1.25 | $10.00 | 85.6 | 8 | n/a | ||
| 13 | Anthropic | 多模态 | 1.0M | $5.00 | $25.00 | 84.6 | 24 | n/a | ||
| 14 | Anthropic | 多模态 | 1.0M | $10.00 | $50.00 | 84.1 | 30 | n/a | ||
| 15 | Anthropic | 多模态 | 1.0M | $30.00 | $150.00 | 83.3 | 12 | n/a | ||
| 16 | Anthropic | 多模态 | 1.0M | $10.00 | $50.00 | 82.0 | 31 | n/a | ||
| 17 | U PaLM 2-L | Unknown | LLM | n/a | n/a | n/a | 81.8 | 4 | n/a | |
| 18 | xiaomi | LLM | 262K | $0.09 | $0.29 | 81.7 | 11 | n/a | ||
| 19 | Anthropic | 多模态 | 1.0M | $2.00 | $10.00 | 81.7 | 23 | n/a | ||
| 20 | OpenAI | 多模态 | 400K | $1.75 | $14.00 | 80.7 | 9 | n/a | ||
| 21 | OpenAI | 多模态 | 1.1M | $30.00 | $180.00 | 79.7 | 12 | n/a | ||
| 22 | Microsoft | LLM | n/a | n/a | n/a | 79.2 | 8 | n/a | ||
| 23 | OpenAI | 多模态 | 1.1M | $2.00 | $10.00 | 78.9 | 27 | n/a | ||
| 24 | OpenAI | 多模态 | 1.1M | $2.00 | $10.00 | 77.9 | 18 | n/a | ||
| 25 | OpenAI | 多模态 | 400K | $1.25 | $10.00 | 77.7 | 8 | n/a | ||
| 26 | xAI | 多模态 | 256K | $1.00 | $2.00 | 77.1 | 8 | n/a | ||
| 27 | Alibaba | LLM | n/a | n/a | n/a | 76.5 | 25 | n/a | ||
| 28 | OpenAI | 多模态 | 1.1M | $2.00 | $12.00 | 75.3 | 31 | n/a | ||
| 29 | OpenAI | LLM | 131K | Free | Free | 74.2 | 10 | n/a | ||
| 30 | DeepSeek | LLM | n/a | n/a | n/a | 73.6 | 15 | n/a | ||
| 31 | Google DeepMind | 多模态 | 1.0M | $1.25 | $10.00 | 73.5 | 4 | n/a | ||
| 32 | Meta | LLM | n/a | n/a | n/a | 73.2 | 11 | n/a | ||
| 33 | nousresearch | LLM | 131K | $0.70 | $0.70 | 73.2 | 6 | n/a | ||
| 34 | minimax | LLM | 205K | $0.30 | $1.20 | 72.4 | 8 | n/a | ||
| 35 | z-ai | LLM | 131K | $0.60 | $2.20 | 72.3 | 7 | n/a | ||
| 36 | Alibaba Qwen | LLM | 33K | $0.66 | $1.00 | 71.4 | 14 | n/a | ||
| 37 | OpenAI | 多模态 | 400K | $21.00 | $168.00 | 71.0 | 7 | n/a | ||
| 38 | Alibaba | LLM | n/a | n/a | n/a | 71.0 | 6 | n/a | ||
| 39 | Anthropic | 多模态 | 1.0M | $5.00 | $25.00 | 70.6 | 29 | n/a | ||
| 40 | Alibaba Qwen | LLM | 1.0M | $1.48 | $4.42 | 70.2 | 24 | n/a | ||
| 41 | Alibaba Qwen | 多模态 | 262K | $0.50 | $3.60 | 69.4 | 29 | n/a | ||
| 42 | Microsoft | LLM | n/a | n/a | n/a | 69.2 | 10 | n/a | ||
| 43 | Anthropic | LLM | n/a | n/a | n/a | 68.7 | 5 | n/a | ||
| 44 | DeepSeek | LLM | 1.0M | $0.09 | $0.18 | 68.6 | 16 | n/a | ||
| 45 | Mistral AI | LLM | 33K | $0.54 | $0.54 | 68.2 | 11 | n/a | ||
| 46 | U PaLM 2-M | Unknown | LLM | n/a | n/a | n/a | 68.1 | 5 | n/a | |
| 47 | xAI | LLM | 131K | $3.00 | $15.00 | 67.9 | 6 | n/a | ||
| 48 | Google DeepMind | LLM | n/a | n/a | n/a | 67.9 | 35 | n/a | ||
| 49 | OpenAI | 多模态 | 400K | $1.75 | $14.00 | 67.6 | 18 | n/a | ||
| 50 | Meta | 多模态 | 1.0M | $1.25 | $4.25 | 67.6 | 6 | n/a | ||
| 51 | Google DeepMind | 多模态 | 1.0M | $0.75 | $3.75 | 67.1 | 18 | n/a | ||
| 52 | Google DeepMind | 多模态 | 1.0M | $0.82 | $4.13 | 66.3 | 30 | n/a | ||
| 53 | DeepSeek | LLM | 1.0M | $1.32 | $3.96 | 66.0 | 21 | n/a | ||
| 54 | moonshotai | 多模态 | 1.0M | $0.24 | $11.94 | 65.8 | 30 | n/a | ||
| 55 | Anthropic | 多模态 | 1.0M | $5.00 | $25.00 | 65.2 | 40 | n/a | ||
| 56 | xAI | 多模态 | 500K | $2.00 | $6.00 | 65.1 | 20 | n/a | ||
| 57 | Alibaba | LLM | n/a | n/a | n/a | 64.9 | 11 | n/a | ||
| 58 | OpenAI | LLM | 400K | $5.00 | $30.00 | 64.9 | 6 | n/a | ||
| 59 | OpenAI | 多模态 | 1.1M | $2.50 | $15.00 | 64.7 | 34 | n/a | ||
| 60 | Anthropic | 多模态 | 1.0M | $0.10 | $0.50 | 64.6 | 25 | n/a | ||
| 61 | z-ai | LLM | 1.0M | $0.03 | $7.00 | 64.3 | 34 | n/a | ||
| 62 | TII | LLM | n/a | n/a | n/a | 64.3 | 4 | n/a | ||
| 63 | Anthropic | 多模态 | 1.0M | $2.00 | $10.00 | 64.3 | 17 | n/a | ||
| 64 | Anthropic | 多模态 | 1.0M | $5.00 | $25.00 | 64.2 | 30 | n/a | ||
| 65 | Alibaba | LLM | n/a | n/a | n/a | 64.1 | 12 | n/a | ||
| 66 | z-ai | LLM | 1.0M | $0.08 | $1.40 | 64.0 | 22 | n/a | ||
| 67 | Meta | LLM | n/a | n/a | n/a | 64.0 | 11 | n/a | ||
| 68 | Google DeepMind | 多模态 | 1.0M | $2.00 | $12.00 | 63.9 | 51 | n/a | ||
| 69 | z-ai | LLM | 205K | $1.00 | $3.20 | 63.8 | 30 | n/a | ||
| 70 | OpenAI | LLM | 8K | $30.00 | $60.00 | 63.7 | 7 | n/a | ||
| 71 | z-ai | LLM | 205K | $1.40 | $4.40 | 63.6 | 26 | n/a | ||
| 72 | Mistral AI | LLM | n/a | n/a | n/a | 63.6 | 12 | n/a | ||
| 73 | Alibaba Qwen | LLM | 82K | $0.20 | $2.40 | 63.5 | 6 | n/a | ||
| 74 | OpenAI | 多模态 | 400K | $0.25 | $2.00 | 63.3 | 8 | n/a | ||
| 75 | moonshotai | 多模态 | 262K | $0.67 | $3.35 | 62.8 | 32 | n/a | ||
| 76 | Alibaba Qwen | 多模态 | 1.0M | $0.33 | $1.95 | 62.5 | 25 | n/a | ||
| 77 | Alibaba Qwen | 多模态 | 262K | $0.25 | $2.00 | 62.0 | 18 | n/a | ||
| 78 | xAI | 多模态 | 500K | $2.00 | $6.00 | 61.7 | 21 | n/a | ||
| 79 | U PaLM 2-S | Unknown | LLM | n/a | n/a | n/a | 61.7 | 5 | n/a | |
| 80 | Microsoft | LLM | n/a | n/a | n/a | 61.5 | 9 | n/a | ||
| 81 | Microsoft | LLM | 66K | $0.62 | $0.62 | 61.4 | 7 | n/a | ||
| 82 | OpenAI | 多模态 | 1.1M | $0.20 | $1.20 | 61.3 | 21 | n/a | ||
| 83 | OpenAI | LLM | 131K | Free | Free | 61.0 | 9 | n/a | ||
| 84 | xAI | 多模态 | 1.0M | $1.25 | $2.50 | 61.0 | 23 | n/a | ||
| 85 | OpenAI | 多模态 | 200K | $20.00 | $80.00 | 60.8 | 10 | n/a | ||
| 86 | DeepSeek | LLM | 1.0M | $0.96 | $1.91 | 60.6 | 34 | n/a | ||
| 87 | Google DeepMind | 多模态 | 1.0M | $1.50 | $9.00 | 60.6 | 26 | n/a | ||
| 88 | U internlm-20b | Unknown | LLM | n/a | n/a | n/a | 60.3 | 7 | n/a | |
| 89 | U Stable Beluga 2 | Unknown | LLM | n/a | n/a | n/a | 60.1 | 13 | n/a | |
| 90 | Alibaba Qwen | LLM | n/a | n/a | n/a | 60.1 | 7 | n/a | ||
| 91 | OpenAI | 多模态 | 400K | $1.25 | $10.00 | 60.1 | 41 | n/a | ||
| 92 | Google DeepMind | 多模态 | 1.0M | $0.75 | $3.75 | 59.9 | 17 | n/a | ||
| 93 | DeepSeek | LLM | 1.0M | $0.44 | $1.32 | 59.8 | 15 | n/a | ||
| 94 | Alibaba Qwen | 多模态 | 1.0M | $2.00 | $6.00 | 59.6 | 24 | n/a | ||
| 95 | Alibaba Qwen | LLM | 33K | $0.36 | $0.40 | 59.6 | 28 | n/a | ||
| 96 | xiaomi | LLM | 1.0M | $1.00 | $3.00 | 59.6 | 13 | n/a | ||
| 97 | U Nemotron-4 15B | Unknown | LLM | n/a | n/a | n/a | 58.9 | 7 | n/a | |
| 98 | DeepSeek | LLM | n/a | n/a | n/a | 58.7 | 15 | n/a | ||
| 99 | DeepSeek | 多模态 | 1.0M | $0.30 | $1.20 | 58.6 | 24 | n/a | ||
| 100 | Google DeepMind | 多模态 | 1.0M | $1.25 | $10.00 | 58.5 | 50 | n/a |
1-100 / 1,269 · 第 1/13 页
…
各类别前 5 名
我们追踪的每种模态中的专项领先者
决策捷径
从排行榜直达对比、提供商、定价和基准路径
常见问题
基于我们数据的快速解答
BenchGecko 追踪多少个 AI 模型?
BenchGecko 目前追踪来自 279 家提供商的 1,269 个 AI 模型,每个模型最多接受 160 项基准测试评分。新模型持续加入,完整数据集每日刷新。
目前最好的 AI 模型是哪个?
“最好”取决于任务。通用推理方面,我们按 3 项以上基准的平均分排名;编程方面,我们专门展示 ELO 和 SWE-bench;性价比方面,我们提供“最便宜的可用模型”指标。你可以使用筛选栏和列排序来选出自己的赢家,也可以从下方的迷你表格中选择一个类别。
平均分是如何计算的?
平均分是模型归一化基准分数的算术平均值,仅在模型至少有一项公开基准结果时计算。基准少于 3 项的模型不会进入领奖台,以避免单一分数造成的异常值。
BenchGecko 的数据来自哪里?
模型元数据和定价来自 OpenRouter 的公开 API。基准数据取自 Epoch AI (CC-BY) 和 SWE-bench 的公开排行榜。ELO 评分来自 LMArena。所有数据每日重新归一化并交叉关联。完整来源请参阅方法论页面。
我可以在文章或产品中使用这些数据吗?
可以。所有 BenchGecko 数据均采用 CC BY 4.0 许可 · 须注明出处。使用上方的“引用此页面”按钮,可获得现成的 APA、MLA、AP Style、BibTeX 和 HTML 嵌入代码。免费 API 层级需要提供指向 benchgecko.ai 的反向链接。
数据多久刷新一次?
核心模型、定价和基准数据每 24 小时刷新一次。当上游来源变化时,实时状态和价格提醒可能更频繁地触发。当上次刷新不到一小时前时,本页的“Live”标签会亮起。
哪些模型是开源的?
开启“开源”筛选,即可只查看提供开源权重的模型。目前我们将 Llama、Qwen、DeepSeek 和 Mistral Open 等宽松 · 开放权重许可的模型计为开源,即使这些权重附带使用限制。
如何并排比较两个模型?
在上方排行榜中勾选任意 2 到 4 个模型,然后点击粉色操作栏中的“比较 →”。你也可以直接访问 /compare/[modelA]-vs-[modelB] 获取可分享的对比页面。