- Published on
LLM 0.37 Adds GPT-6.1 Models, Endpoint Command and Smarter Embeddings
- Authors
- Name
- DeepHorus
LLM, a command-line tool and Python library, released version 0.37 on November 2, 2026. The update adds two GPT-6.1 models, gpt-6.1-sol and gpt-6.1-terra, and a new endpoint command. It also changes the default model and improves embeddings.
The release represents about two months of accumulated features and fixes. Users without a personal default now get gpt-6-luna, which costs half as much as the previous default, gpt-5.6-luna. This aims to reduce costs.
A key addition is the llm openai endpoint add command, which simplifies configuring models that use OpenAI-compatible endpoints. It makes connecting third-party services easier. The command streamlines setup for custom deployments.
Embeddings now accept different content types for different binary content. This is needed for models like Gemini Embedding 2, which can embed text, audio, images, and video. The change enables flexible multimodal workflows.
The release also touches on AI-assisted pull requests. While many maintainers complain about a flood of such contributions, LLM's maintainer says it has not been a problem. The maintainer uses Opus 5.5 or GPT-6 Astra to help triage submissions.
This release includes fixes from 12 authors, almost all AI-assisted. The maintainer treats them as bug reports with attached examples, sometimes landing them and sometimes writing a different fix. Contributors thanked include ac1982, Allen Flux, Christopher Krah, and nosmile99, among others.
The LLM 0.37 update reflects continued evolution over two months. By adding models, simplifying endpoint configuration, and improving embeddings, it aims to keep pace with changing LLM capabilities. The maintainer's positive experience with AI-assisted contributions may offer a perspective for other projects.
