IBM发布Granite 4.2：为开放企业模型带来原生推理和代理强化学习
IBM Releases Granite 4.2: Bringing Native Reasoning and Agentic RL to Open Enterprise Models
IBM发布了带有内置代理能力的开放式Granite 4.2系列，并在Apache 2.0下授权
IBM has released the Granite 4.2 series of language models, which come in three sizes: 3B, 8B, and 30B, trained on approximately 1.5 trillion tokens, with a maximum context window of up to 512,000 tokens. The larger models were trained using "proxy RL" and are able to learn tool use and code execution autonomously. All models are released under the Apache 2.0 license.
Verify source →IBM发布Granite 4.2：为开放企业模型带来原生推理和代理强化学习
IBM released Granite 4.2, an open family of reasoning language models in three sizes: 3B, 8B, and 30B, all under the Apache 2.0 license. All models support native tool calling and thinking/low-effort/non-thinking switching. The 8B and 30B models were also trained with agent RL and can edit code, drive terminals and run web searches in a sandbox environment.
Verify source →