IBM Unveils Granite Switch 4.1: One-Endpoint Chat, Decision, and Speech Model
Current overview (not historical)
IBM has released the Granite Switch 4.1 series, a single‑endpoint model that combines natural‑language chat, professional decision‑making, and speech‑to‑text transcription. By integrating Decision aLoRA with the fastest ASR engine, Granite‑Speech‑5.0‑Turbo‑CTC, the new model merges three distinct technologies into one system. Its memory footprint is roughly equal to running the base model alone, thanks to a shared KV cache that makes adding expert adapters virtually cost‑free. In a real‑world five‑step Retrieval‑Augmented Generation (RAG) workflow, Granite‑Switch‑4.1‑30B outperformed competing models in speed, demonstrating the practical benefits of this unified approach.
Written with AI from the reporting · updated 10/29/2026
Expand the detailed fact route and sources
Story route
A station-by-station route from the first public evidence onward; every station carries its own evidence and uncertainty.
- 1Unclassified fact stationOct 16, 00:00recorded Oct 16, 04:35single sourcerevised
Granite Switch 4.1 发布，实现多模态任务切换
Read original source ↗ Hugging Face BlogReplay this moment in the explorerEvidence & revisions · 1 source records
- Subject：
- ibm-granite
- Action：
- 发布 Granite Switch 4.1 系列模型
- Object：
- 集成决策适配器和语音转文本模型的统一端点
- Occurred (fact side)：
- Oct 16, 00:00
- Published (valid side)：
- Oct 16, 04:34
- First seen (known side)：
- Oct 16, 04:35
Fact stationSingle source · one-sided claimRevisedCopy may have been revised later; time replay is bounded. Layer labels group reading only, not causation between layers.
- 2Unclassified fact stationOct 29, 22:29single sourcereport-time onlyrevised
Granite Switch 4.1.1 更新模型如更新软件
Read original source ↗ Hugging Face BlogReplay this moment in the explorerEvidence & revisions · 1 source records
- Subject：
- hugging-face
- Action：
- 发布补丁版本
- Object：
- Granite Switch 4.1.1
- Occurred (fact side)：
- not recorded
- Published (valid side)：
- Oct 29, 21:37
- First seen (known side)：
- Oct 29, 22:29
Fact stationSingle source · one-sided claimRevisedReport-time onlyCopy may have been revised later; time replay is bounded. Layer labels group reading only, not causation between layers.
Uncertainty marks flag where to check the sources; they are not proven causation.
Report timeline
Explore every side through dated source reports.
- Hugging Face BlogFirst-party sourceRelated fact: single sourceFact occurrence time unverifiedSource report revisedHugging Face releases Granite Switch 4.1.1 update with multilingual ASR model and improved Decision adapter
Report first recorded Oct 29, 22:29
Hugging Face released Granite Switch 4.1.1, a backwards-compatible patch update that adds a multilingual Automatic Speech Recognition model supporting 10 languages with native punctuation and capitalization, and an improved Decision adapter in granitelib-core showing better performance on the Jev Decision Index test set. The update follows the software-style versioning model where the base model remains untouched and only adapters are updated.
Read original source ↗
- Hugging Face BlogFirst-party sourceRelated fact: single sourceSource report revisedHugging Face releases Granite Switch 4.1 enabling chat, decision-making, and speech from one endpoint
Report first recorded Oct 16, 04:35
Granite Switch 4.1 models, including Granite-Switch-4.1-8B and -30B, integrate a Decision aLoRA for general decision-making, a specialized Guardian aLoRA that outperforms Jev 1.13, Microsoft-Decision-1, GPT-Luna-Decisions, and Clef-Flash on safety tasks, and Granite-Speech-5.0-Turbo-CTC, the fastest ASR model on the OpenASR Leaderboard. All these capabilities run from a single hosted endpoint, using approximately the same VRAM as the base model alone, allowing for fast switching between chat, decision, and spoken requests.
Read original source ↗
Event activity trend
- Latest comparable
- 3
- Comparable peak
- 10Oct 29, 23:00
- 24h change
- -51%
The chart compares consistently observed participants only. Hover or tap to inspect each hour; use arrow keys to move between observations.
Event discussion
Reader opinions are hosted on GitHub Discussions; posting requires a GitHub account. Comments are not verified facts.