What happened
IBM released its Granite 4.2 language models in 3B, 8B, and 30B sizes. The models were trained from scratch on about 15 trillion tokens and support context windows up to 512,000 tokens.
Why it matters
The 8B and 30B variants undergo agentic RL training to use tools, write and run code, and search the web in real sandbox environments. The 30B model leads across all three agentic benchmarks, and all models run on vLLM or SGLang.
What to watch
IBM also introduced Granite Speech 5.0 Turbo CTC models with just 470 million parameters. They are twice as fast as the previous leaders on the Open ASR Leaderboard and can transcribe three hours of audio in one second.
Summaries like this, in your inbox every morning.
IBM's release of Granite 4.2 builds on the trend of open-weight models that aim to offer competitive performance in agentic workflows. The integration of agentic RL training in the 8B and 30B variants directly addresses the growing need for models that can interact with external tools and execute tasks autonomously. By supporting OpenAI-format tool calling and running on popular inference engines like vLLM or SGLang, IBM positions these models as practical options for developers integrating AI into production systems.
The availability of the models under the Apache 2.0 license on multiple platforms lowers the barrier for adoption, particularly for businesses looking to avoid proprietary lock-in. The Granite Speech 5.0 Turbo models, with their speed advantage on the Open ASR Leaderboard, complement the text models, potentially offering a comprehensive on-premise or hybrid AI toolset. As the landscape of open-weight models grows, performance on specialized tasks like agentic tool use will likely become a key differentiator for enterprises making deployment decisions.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Chunghwa Telecom president Rong-shy Lin said the shift from conversational AI to autonomous AI agents is resha…
Apple and Meta said they would jointly release a new open-weight AI model designed to run on high-end Macs, ai…
A writer and entrepreneur opened his daily to-do list and found 156 items, and concluded that an AI assistant…
A Zenn contributor sorted AI use into three modes — AI as an assistant, AI running alongside people, and AI th…
A non-engineer operator says one person ships four smartphone apps on Google Play, three on the App Store, wit…
The walkthrough points the CLI at a local mock via the ANTHROPIC_BASE_URL environment variable
