Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint

Back to Articles Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint Enterprise Article Published October 15, 2026 Upvote 8 +2 Kate Soule katesoule Follow ibm-granite Subhajit Chaudhury subhajit1412 Follow ibm-granite Hendrik Strobelt hen Follow ibm-granite Luis A Lastras GrizleeBer Follow ibm-granite

回到文章 Granite Switch 4.1：从单一端点实现聊天、决策和语音。企业文章，发布于 2026 年 10 月 15 日。点赞 8 +2。作者：Kate Soule, Subhajit Chaudhury, Hendrik Strobelt, Luis A Lastras。

Introduced last spring, the Granite Switch project enables a modular approach to model development. Now Granite Switch 4.1 models, armed with a Decision adapter and the fastest ASR model on the OpenASR leaderboard, can move efficiently between natural-language responses, general and specialized decisions, and speech-to-text tasks—all from one hosted endpoint.

Granite Switch 项目于去年春天推出，实现了模型开发的模块化方法。现在，配备了决策适配器 (Decision adapter) 和 OpenASR 排行榜上最快 ASR 模型的 Granite Switch 4.1 模型，可以在自然语言响应、通用及专业决策以及语音转文字任务之间高效切换——所有这些都通过一个托管端点完成。

Specifically, Granite-Switch-4.1-8B and -30B models are configured to include: A new Decision aLoRA for general-purpose decision-making (e.g. v1/systemone), A specialized Guardian aLoRA that, on average, outperforms Jev 1.13, Microsoft-Decision-1, GPT-Luna-Decisions, and Clef-Flash on safety-related decision tasks, and Granite-Speech-5.0-Turbo-CTC, the fastest ASR model on the OpenASR Leaderboard, embedded for use in transcription tasks.

具体而言，Granite-Switch-4.1-8B 和 -30B 模型配置包括：用于通用决策的新型 Decision aLoRA（例如 v1/systemone）；在安全相关决策任务中平均表现优于 Jev 1.13、Microsoft-Decision-1、GPT-Luna-Decisions 和 Clef-Flash 的专业 Guardian aLoRA；以及嵌入用于转录任务的 Granite-Speech-5.0-Turbo-CTC（OpenASR 排行榜上最快的 ASR 模型）。

All three of these technologies (along with the underlying base model) are composed into one model, taking up approximately the same amount of VRAM on a GPU as the base model would by itself. That means more capabilities with fewer GPUs. And because the Switch adapters were built as cache-efficient Activated LoRAs, the KV cache created from a base model generation can be reused for any subsequent adapter-based decisions.

这三项技术（连同底层基础模型）被组合成一个模型，在 GPU 上占用的显存量与基础模型本身大致相同。这意味着可以用更少的 GPU 实现更多的功能。由于 Switch 适配器是作为缓存高效的激活 LoRA (Activated LoRAs) 构建的，因此从基础模型生成中创建的 KV 缓存可以重复用于任何后续基于适配器的决策。

Combined with our team’s recent vLLM contributions to speed up low rank adapter inferencing, as well as a co-located ASR encoder, Granite Switch 4.1 models can switch between a chat request and a decision request and a spoken request, with blazing fast speeds. Check out our Speech-to-Decisions HF space where granite-switch-4.1-8b transcribes audio, generates summaries, screens for harmful content, and tests 15 other user-defined flags—all running on a single endpoint in real time.

结合我们团队近期为加速低秩适配器推理所做的 vLLM 贡献，以及同地部署的 ASR 编码器，Granite Switch 4.1 模型可以在聊天请求、决策请求和语音请求之间以极快的速度切换。请查看我们的 Speech-to-Decisions HF 空间，其中的 granite-switch-4.1-8b 模型可以实时转录音频、生成摘要、筛选有害内容并测试其他 15 个用户定义的标志——所有这些都在单个端点上运行。

Granite Libraries: The OG Decision Models. Decision models aren’t new. Our previous release of Granite Libraries features a repository of specialized “decision model” adapters and other specialized task adapters that can be composed into custom Granite Switch models. And while Jev and other closed decision models focus on generality, we’ve found that many of our specialized granitelib adapters can outperform generalists on specialized classification tasks.

Granite 库：原始决策模型。决策模型并不新鲜。我们之前发布的 Granite 库包含一个专门的“决策模型”适配器和其他专业任务适配器的存储库，这些适配器可以组合成自定义的 Granite Switch 模型。虽然 Jev 和其他封闭式决策模型侧重于通用性，但我们发现我们的许多专业 granitelib 适配器在专业分类任务上可以胜过通用模型。

Compare granitelib adapters with generalist decision models when evaluated on specialized classification tasks. Because adapters share the base model’s KV cache, adding specialists alongside a generalist decision adapter costs almost nothing. You get broad coverage from the generalist and higher accuracy from specialists on the tasks you care about.

在专业分类任务中评估时，请比较 granitelib 适配器与通用决策模型。由于适配器共享基础模型的 KV 缓存，在通用决策适配器旁边添加专业适配器几乎不需要额外成本。您可以从通用模型获得广泛的覆盖范围，并从专业模型在您关心的任务上获得更高的准确性。

In a “Race” between Granite-Switch-4.1-30B (frozen base Granite + 4 specialized aLoRAs) and Jev-27B (frozen base Qwen model + 1 generic decision LoRA), Granite-Switch-4.1-30B was able to complete a real world, 5-step RAG workflow much faster than its competitor.

在 Granite-Switch-4.1-30B（冻结的基础 Granite + 4 个专业 aLoRA）与 Jev-27B（冻结的基础 Qwen 模型 + 1 个通用决策 LoRA）的“竞赛”中，Granite-Switch-4.1-30B 完成现实世界 5 步 RAG 工作流的速度远快于其竞争对手。

Compose and Extend Your Own Granite Switch Model. Granite-Switch-4.1-8B and -30B are released with “batteries included”. They are Granite Switch models that we compiled with a predefined set of helpful adapters. But you don’t need to use our version. Anyone can compose their own Granite Switch model, mixing and matching adapters from Granite Libraries to best meet their needs.

组合并扩展您自己的 Granite Switch 模型。Granite-Switch-4.1-8B 和 -30B 发布时“自带电池”。它们是我们使用预定义的一组有用适配器编译的 Granite Switch 模型。但您不必使用我们的版本。任何人都可以组合自己的 Granite Switch 模型，混合并匹配 Granite 库中的适配器，以最好地满足其需求。

Moving forward, we see tremendous opportunity for the community to compose and extend Granite Switch models with custom and third party adapters. We plan to release an Adapter SDK for training adapters in Mellea, along side other tutorials and Granite Switch support. These adapters could be specialized classification adapters to augment decision-based workflows or adapters for specialized language tasks.

展望未来，我们认为社区在利用自定义和第三方适配器组合及扩展 Granite Switch 模型方面拥有巨大的机会。我们计划发布一个用于在 Mellea 中训练适配器的适配器 SDK，以及其他教程和 Granite Switch 支持。这些适配器可以是用于增强基于决策的工作流的专业分类适配器，也可以是用于专业语言任务的适配器。

You could even use custom adapters to teach a 3B parameter version of Granite to play the classic game Doom. Similarly to Jev, our custom 3B parameter Granite Switch model made split second decisions to navigate the Doom game, but was able to do so at 60 - 100 decisions / second, compared to Jev’s reported 10 decisions / second. Unlike Jev, this Granite Switch model can also narrate and respond to your questions while it is playing.

您甚至可以使用自定义适配器来教 3B 参数版本的 Granite 玩经典游戏《毁灭战士》(Doom)。与 Jev 类似，我们自定义的 3B 参数 Granite Switch 模型能够做出瞬间决策来导航《毁灭战士》游戏，但其速度达到了每秒 60 - 100 次决策，而 Jev 的报告速度为每秒 10 次决策。与 Jev 不同的是，这款 Granite Switch 模型在玩游戏时还可以进行解说并回答您的问题。

We look forward to releasing more details and data on Granite Switch in the upcoming weeks, but until then, we encourage you to try out our latest Switch models and these other helpful resources for getting started.

我们期待在未来几周内发布关于 Granite Switch 的更多详细信息和数据，在此之前，我们鼓励您尝试我们最新的 Switch 模型以及这些其他有用的入门资源。

Resources for Getting Started: Granite Switch 4.1 Model Cards, Speech-to-Decisions Demo, Granite Switch GitHub Repo, Original Granite Switch and Granite Libraries Release Blog, Create your own aLoRA v LoRA race on Colab.

入门资源：Granite Switch 4.1 模型卡、Speech-to-Decisions 演示、Granite Switch GitHub 仓库、原始 Granite Switch 和 Granite 库发布博客、在 Colab 上创建您自己的 aLoRA 与 LoRA 竞赛。