LFM3
A model newly added to the Ollama MLX back end, capable of running through the MLX path on Apple Silicon devices. Specific parameters and capabilities have not been disclosed.
Timeline (last 90 days)
The release notes do not give specific performance data or hardware requirements for LFM3 on Linux/Windows.
Previously, LFM3 on Ollama could only run on Apple silicon devices.
LFM is Liquid AI's foundation model product line, positioned as a lightweight, efficient, locally deployable general-purpose model family; LFM3 is the third-generation version of the series, with significantly expanded multimodal capabilities compared with its predecessors.
LFM3 is a model series launched by Liquid AI, and its lfm3-audio and lfm3-omni variants support text-to-speech capability.
In Ollama v0.40.13, the supported platforms for the LFM3 series models expanded from Apple silicon only to Linux and Windows.
The 1.7B text model can be fine-tuned on a single GPU.
Currently the public information focuses on model specifications and deployment conditions, and the official side has not yet given the specific capability boundaries of each modality, actual inference speed, or horizontally comparable data on performance across different hardware.
The six LFM3 models use open weights, share the same backbone network, and cover text, vision, audio, omni-modal and decision models.
The LFM3 product on Product Hunt received 104 votes, ranked 5th for the day, and had 3 comments in the comment section.
Liquid AI positions LFM3 for three types of edge devices — phones, laptops and robots — and the goal is not cloud inference.
15 more timeline events
All Facts (20)
LFM is Liquid AI's foundation model product line, positioned as a lightweight, efficient, locally deployable general-purpose model family; LFM3 is the third-generation version of the series, with significantly expanded multimodal capabilities compared with its predecessors.
50%UnverifiedLFM3 is a model series launched by Liquid AI, and its lfm3-audio and lfm3-omni variants support text-to-speech capability.
50%UnverifiedPreviously, LFM3 on Ollama could only run on Apple silicon devices.
50%UnverifiedThe release notes do not give specific performance data or hardware requirements for LFM3 on Linux/Windows.
50%UnverifiedLiquid AI positions LFM3 for three types of edge devices — phones, laptops and robots — and the goal is not cloud inference.
50%UnverifiedThe LFM3 model weights are already available on Hugging Face.
50%UnverifiedThe 1.7B text model can be fine-tuned on a single GPU.
50%UnverifiedThe six LFM3 models use open weights, share the same backbone network, and cover text, vision, audio, omni-modal and decision models.
50%UnverifiedCurrently the public information focuses on model specifications and deployment conditions, and the official side has not yet given the specific capability boundaries of each modality, actual inference speed, or horizontally comparable data on performance across different hardware.
50%UnverifiedThe LFM3 product on Product Hunt received 104 votes, ranked 5th for the day, and had 3 comments in the comment section.
50%UnverifiedThe original material on LFM3 does not cover training data composition, benchmark results, or license details.
50%UnverifiedThe six models in the LFM3 series cover directions including text, vision, audio, omni, and decisions.
50%UnverifiedLFM3-1.7B adopts a tri-hybrid backbone structure, supports a 256K context window, and is about 1GB in size under 4-bit quantization.
50%UnverifiedThere is currently no public benchmark data showing the performance of LFM3 and d1 on MLX.
50%UnverifiedLFM3 emphasizes "Specialized Intelligence", that is, making the model genuinely usable in specific domains and on specific tasks, rather than pursuing all-round dominance in general capabilities.
50%UnverifiedThe LFM3 technical details officially disclosed so far are limited, and which modalities are specifically covered and the performance level of each modality still await empirical verification.
50%UnverifiedThe core claim of LFM3 is "Every Modality", meaning that beyond text it also covers input and output capabilities in other modalities.
50%UnverifiedCommunity discussion on Hacker News shows that developers care more about LFM3's actual usable interfaces, degree of open source, and real-task performance than about its marketing language.
50%UnverifiedOn Hacker News, the LFM3 launch post received 17 points and 6 comments.
50%UnverifiedAt the time the article was published, the original release information did not disclose the specific parameter scales, architecture details or performance data of LFM3 and d1.
50%