Ollama v0.40.12 Released with Liquid AI LFM3 Model Family Support

Liquid AI's LFM3 model family represents a significant step forward for local LLM deployment, offering purpose-built models for on-device inference with impressive capabilities at minimal footprint. The 1.7B base model, complemented by vision-enabled (LFM3-VL) and speech-enabled (LFM3-Audio) variants, all operate with a 256K token context window—enabling sophisticated multi-turn conversations and complex reasoning tasks without cloud infrastructure. Ollama's native integration ensures these models are accessible to practitioners through a single command.

The tri-hybrid architecture underlying LFM3 is engineered specifically for local deployment constraints: optimized for tool calling, structured output, and agentic workflows that have previously required larger models or API calls. With v0.40.12 also addressing prompt speed and small-image detail quality, this release removes practical barriers to building production-grade local AI systems. For teams prioritizing data privacy, latency, and operational autonomy, LFM3 via Ollama provides a compelling alternative to cloud-dependent solutions.

Read the full article on Ollama release.

Source: Ollama release · Relevance: 9/10