IBM 发布开源 Granite 4.2 系列，内置智能体能力并采用 Apache 2.0 许可

IBM drops open-weight Granite 4.2 family with built-in agentic capabilities under Apache 2.0

AI 导读

IBM 发布 Granite 4.2 语言模型，提供 3B、8B 和 30B 三种尺寸，基于约 15 万亿 token 从头训练，支持最高 512,000 token 的上下文窗口，并可在“思考”与“非思考”模式间切换以控制每项任务的计算量。

正文 · 原文

IBM has released its Granite 4.2 language models in 3B, 8B, and 30B sizes. Trained from scratch on about 15 trillion tokens, the models support context windows up to 512,000 tokens and can toggle between "thinking" and "non-thinking" modes to control compute per task, according to IBM. A "low-effort" mode saves resources on simple queries. The 8B and 30B variants also go through what IBM calls "agentic RL" training, where they learn to use tools, write and run code, and search the web in real sandbox environments. All models support OpenAI-format tool calling and run on vLLM or SGLang.

Benchmark results for the three Granite 4.2 models on agentic tasks and tool use. The 30B model leads across all three tests. | Image: IBM

The new Granite Speech 5.0 Turbo CTC models have just 470 million parameters and are twice as fast as the previous leaders on the Open ASR Leaderboard, IBM says. They can transcribe three hours of audio in one second. All models are available under the Apache 2.0 license on Hugging Face, Ollama, GitHub, and other platforms.

IBM