LMStudio

LM Studio announced LM Studio Bionic on July 16, 2026, describing it as the company's "biggest leap forward" — an AI agent built specifically for open models that handles coding, research, and complex work with documents and files. The product supports flexible model execution: running locally, connecting through LM Li The launch bundles three Bionic capabilities. Offline voice transcription ships at launch with Voxtral by Mistral AI as a multilingual realtime transcription model for the voice keyboard. The coding mode inspects local codebases, explains unfamiliar code, and edits/debugs with inline diffs and agentic code search, work

LM Studio's official blog index shows a dense cadence of first-party technical posts through summer 2026. Recent entries include Splash Engine from Inco AI for the fastest local Qwen3.8 on Apple Silicon (Sep 18), Session References and Introspection in Bionic (Sep 17), Auto Review in Bionic covering AST parsing and saf Earlier items in the same index add more provider-level capabilities: DeepSeek V4 Flash available both locally and in LM Studio Bionic on US-hosted servers with Zero Data Retention by default (Aug 2), Kimi K3 availability in Bionic (Jul 27), the original Bionic launch (Jul 16), MLX Engine v1.8.5 with KV-cache checkpoin

Stable release notes and version-by-version product updates for LM Studio.