LFM2-24B-A2B

Good fit for

People building local document-question-answering or summarization workflows with external retrieval tools.

Why consider it

The exact provider card documents a non-reasoning instruct model, a 32,768-token context and supported local inference frameworks.

Main limitations

The provider explicitly does not recommend this model for coding, so Coding is unassigned. Approximately 2B active parameters do not mean only 2B parameters must be stored; detailed provider metadata gives 2.3B active. Web-search examples require externally executed tools, so no built-in Web Search or File Uploads feature is assigned.

Cost and access

No hosted subscription or token fee for local inference within the license’s permitted uses. Hardware, electricity and setup are separate. LFM Open License v1.0 applies. Review the complete custom license before production; do not assume Apache or MIT.

Getting started

Install a current compatible Ollama version, review the exact model license, then run: ollama run lfm2:24b. API means the local runtime endpoint, not a hosted service included with the weights.