Granite Switch 4.1: chat, decisions and speech from one endpoint
On October 15, 2026, IBM published Granite Switch 4.1, in 8B and 30B, combining chat, decision and safety adapters and speech recognition with about the same GPU memory as the base model.
According to IBM, Granite-Switch-4.1-8B and -30B include a general-purpose decision adapter, a Guardian adapter for safety decisions and Granite-Speech-5.0-Turbo-CTC, which the company describes as the fastest speech recognition model on the OpenASR leaderboard. The adapters reuse the base model's cache, so one endpoint switches between chat, classification and audio.
For an organization, this may mean less GPU to serve voice, content screening and support. IBM has not published Portuguese numbers and says more data is coming in the next weeks, so test with real audio and text from your domain before adopting it.