Hugging Face 热门模型第 14 名：mistralai/Mistral-Large-4-1T-A52B，1208 赞，近期下载 278 次。
标签：transformers, mistral_large4, image-text-to-text, vLLM, SGLang, llama.cpp, conversational, en, fr, de, es, pt
README：
Mistral Large 4 - Le Chonk
Mistral Large 4 是我们迄今为止最大、能力最强的模型：一个拥有 1 万亿参数的原生多模态混合专家（MoE）模型，每个 token 有 490 亿活跃参数（包括嵌入层和输出层共 520 亿），在 Mistral 位于欧洲的数据中心中，使用 3,800 块 NVIDIA Grace Blackwell GPU 从头训练而成。
Mistral Large 4 将指令遵循、推理、编码和智能体工作流统一在一组权重中。虽然它是一个通用模型，但在网络安全、金融和法律等工作负载上表现出特别优异的性能，并且在视觉定位方面超越了前沿的闭源模型。它原生支持 160 多种语言，包括欧盟的所有官方语言。
更多信息请访问我们的 博客。该模型也通过 Mistral API 提供服务，API 中的模型名称为 mistral-large-4。
[!Note]
检查点包含一个内嵌的 DSpark 推测解码草稿头 —— 请参阅 推测解码 (DSpark)。
[!Note]
还提供量化后的 NVFP4 检查点，具有更小的内存占用、更快的推理速度以及相同的性能。它针对 Blackwell GPU 进行了优化，并支持 vLLM 和 SGLang。
关键特性
Mistral Large 4 包含以下架构选择：
- ~1T 总参数 / ~52B 活跃参数：细粒度的混合专家结构，包含 128 个路由专家（每个 token 激活 4 个）和 1 个共享专家。
- 1M 上下文长度，适用于长上下文工作负载。
- 多模态输入：接受文本和图片输入，并输出文本。采用 Pixtral 风格的视觉编码器，能够处理可变尺寸和纵横比的图像。
- 指令和推理功能：支持函数调用，可根据请求配置推理力度。
Mistral Large 4 提供以下能力：
- 推理模式：三种推理力度 ——
low（低）、medium（中）、high（高，默认）—— 使得同一模型既能回答快速的聊天回复，也能处理复杂的智能体运行任务。
- 视觉：分析图片、图表、技术图纸和文档，在开源权重模型中具有最先进的视觉定位能力。
- 多语言：原生支持 160 多种语言，包括欧盟的所有官方语言。
- 系统提示词：对系统提示词具有强大的遵循和支持能力。
- 智能体：原生支持函数调用、JSON 输出以及严格的 JSON Schema 强制约束。
- 网络安全：在安全工作中处于开源权重模型的最前沿，同时会拒绝恶意请求。
- 大上下文窗口：支持 1M 上下文窗口。
我们根据 Mistral 生态系统开源权重许可证 发布此模型：这是 MIT 许可证的修改版本，适用于商业和非商业用途 —— 运营 Model-as-a-Service 业务需要与 Mistral 签订单独的商业许可证。
推荐设置
- 推理力度：
'low' / 'medium' / 'high' —— 默认为 'high'。- 对于复杂提示词和智能体用途，请使用
reasoning_effort="high"；对于快速聊天回复使用 'low'，对于日常使用使用 'medium'。
- 采样：对于所有推理力度，我们推荐以下采样参数：
temperature=1.0top_p=0.95
- 系统提示词：定义清晰的环境和用例，包括如何有效利用智能体系统中工具的指导。
- 工具：保持工具集定义明确，并将数量限制为用例所需的最小值。
- 上下文：检查点支持 1M 上下文窗口。
基准测试
我们将 Mistral Large 4 与最近的开源权重模型进行了比较。此处报告了智能体编码和全球智能的结果；更多基准测试结果，特别是在网络安全、金融和法律领域，请参阅我们的 博客文章。
智能体编码
Mistral Large 4 以前沿的编码性能著称，同时显著减少了 token 的使用量，使得在大规模运行时，复杂的编码智能体更具经济性。
在 Artificial Analysis Coding Agent Index 上，ML4 得分 48%，与领先的开源权重模型 GLM-5.3 持平，同时在每个任务上平均使用的 token 数减少了 47%。
全球智能
在 Artificial Analysis Intelligence Index v5 上，Mistral Large 4 得分 26，是西方开源权重模型中最高的。来自中国的领先开源权重模型在该综合指数上得分更高，但如我们的 博客文章 所示，Mistral Large 4 在编码和网络安全方面表现尤为强劲。
演示画廊
Le Chonk Gallery 是一个实时演示空间画廊，展示了由我们和社区使用 Mistral Large 4 构建的 Demo Spaces，涵盖编码、视觉与定位、工具调用、JSON Schema、多语言等。任何标记为 Le_chonk 的空间都会出现在这里。
官方演示在 Mistral API 上运行，在发布的第一周，每位访客享有少量共享免费额度，因此您可以免费试用该模型。您也可以使用自己的 Mistral API 密钥。
要构建演示，我们推荐使用 Mistral Vibe，这是由 Mistral Large 4 驱动的智能体编码工具。
使用方法
您可以在多个用于推理和微调的库中找到对 Mistral Large 4 的支持。
我们感谢所有生态系统的伙伴，他们在此次发布过程中帮助我们整合 Le
查看原文
Hugging Face 热门模型第 14 名：mistralai/Mistral-Large-4-1T-A52B，1208 赞，近期下载 278 次。
标签：transformers, mistral_large4, image-text-to-text, vLLM, SGLang, llama.cpp, conversational, en, fr, de, es, pt
README：
Mistral Large 4 - Le Chonk
Mistral Large 4 is our largest and most capable model to date: a 1 trillion-parameter natively multimodal Mixture-of-Experts model with 49 billion active parameters per token (52 billion including embeddings and output layers), trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own datacenters in Europe.
Mistral Large 4 unifies instruction following, reasoning, coding, and agentic workflows in a single set of weights. While a generalist model, it shows especially good performance on workloads such as cybersecurity, finance, and law, and it surpasses frontier closed models on visual grounding. It is natively fluent in 160+ languages, including every official language of the European Union.
Find more information on our blog. The model is also served on the Mistral API as mistral-large-4.
[!Note]
The checkpoint ships with an embedded DSpark speculative decoding draft head — see Speculative decoding (DSpark).
[!Note]
A quantized NVFP4 checkpoint is also available, with a smaller memory footprint, faster inference and same performance. It targets Blackwell GPUs and supports vLLM and SGLang.
Key Features
Mistral Large 4 includes the following architectural choices:
- ~1T total / ~52B active parameters: granular Mixture-of-Experts with 128 routed experts (4 active per token) and 1 shared expert.
- 1M context length, for long-context workloads.
- Multimodal input: accepts text and image input, with text output. Pixtral-style vision encoder trained to handle variable image sizes and aspect ratios.
- Instruct and Reasoning functionalities with function calls, reasoning effort configurable per request.
Mistral Large 4 offers the following capabilities:
- Reasoning Mode: three reasoning efforts —
low, medium, high (default) — so the same model can answer a quick chat reply or work through a complex agentic run.
- Vision: analyzes images, charts, technical drawings, and documents, with state-of-the-art visual grounding among open weights.
- Multilingual: natively fluent in 160+ languages, including all official languages of the European Union.
- System Prompt: strong adherence and support for system prompts.
- Agentic: native function calling, JSON output, and strict JSON schema enforcement.
- Cybersecurity: state of the art among open-weight models for security work, while refusing malicious requests.
- Large Context Window: supports a 1M context window.
We release this model under the Mistral Ecosystem Open-Weight License: a modified version of the MIT license, for both commercial and non-commercial use — operating a Model-as-a-Service business requires a separate commercial license with Mistral.
Recommended Settings
- Reasoning Effort:
'low' / 'medium' / 'high' — default is 'high'.- Use
reasoning_effort="high" for complex prompts and agentic usage, 'low' for fast chat replies and 'medium' for day to day usage.
- Sampling: for all reasoning efforts, we recommend the following sampling parameters:
temperature=1.0top_p=0.95
- System Prompt: define a clear environment and use case, including guidance on how to effectively leverage tools in agentic systems.
- Tools: keep the set of tools well-defined and limit their number to the minimum required for the use case.
- Context: the checkpoint supports a 1M context window.
Benchmarks
We compared Mistral Large 4 with recent open-weight models. We report results on agentic coding and global intelligence here; for more benchmarks, especially on cybersecurity, finance and law, see our blog post.
Agentic Coding
Mistral Large 4 delivers frontier coding performance with significantly fewer tokens, making sophisticated coding agents more economical to run at scale.
On the Artificial Analysis Coding Agent Index, ML4 scores 48%, effectively tied with the leading open-weight model, GLM-5.3, while using 47% fewer tokens per task on average.
Global Intelligence
On the Artificial Analysis Intelligence Index v5, Mistral Large 4 scores 26, the highest among western open-weight models. The leading open-weight models from China score higher on this aggregate index, but Mistral Large 4 is particularly strong on coding and cybersecurity, as shown in our blog post.
Demo Gallery
Le Chonk Gallery is a live gallery of demo Spaces built with Mistral Large 4, from us and from the community: coding, vision and grounding, tool calls, JSON schema, multilingual, and more. Any Space tagged Le_chonk appears there.
Official demos run on the Mistral API with a small shared free budget per visitor for the first week of the release, so you can try the model without paying. You can also use your own Mistral API key.
To build a demo, we recommend Mistral Vibe, our coding agent powered by Mistral Large 4.
Usage
You can find Mistral Large 4 support on multiple libraries for inference and fine-tuning.
We thank all the ecosystem folks who helped us throughout this release to integrate Le
原文版权归 Hugging Face 热门模型 及原作者所有，本站转载仅供阅读。如需删除，请联系 support@aisa.one。
Studio