Simon Willison Updates LLM CLI Tool And Python Library
Simon Willison releases a new version of his LLM CLI tool and Python library, featuring accumulated updates and fixes. The update includes new models and improved support for embeddings.
How this was written. Content Delta wrote this article with a language model from Simon Willison’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.
Key points from Simon Willison’s report
- Simon Willison has released a new version of his LLM CLI tool and Python library, featuring accumulated updates and fixes.
- The update includes new models, including gpt-6.1-sol and gpt-6.1-terra, and improved support for embeddings.
- The default model has been switched to gpt-6-luna, which is half the price of the previous default model.
- The LLM tool and library now support embeddings with different content types.
- Simon Willison's project has been affected by a flood of AI-assisted pull requests, but he has found this to be helpful.
Simon Willison has released a new version of his LLM CLI tool and Python library, according to Simon Willison. This release represents around two months of accumulated features and fixes. The update includes several key additions, including new models and improved support for embeddings. The new release is a significant update to the tool and library.
The new release includes the addition of gpt-6.1-sol and gpt-6.1-terra models, as well as a new command to make it easier to configure additional models that use OpenAI-compatible endpoints. This update provides users with more options and flexibility when working with the tool and library. The new command is designed to simplify the process of adding new models.
Simon Willison has also switched the default model to gpt-6-luna, which is half the price of the previous default model, gpt-5.6-luna. This change is expected to make the tool and library more accessible to users. The new default model is designed to provide a more cost-effective option for users.
The LLM tool and library now support embeddings with different content types, which is necessary for models like Gemini Embedding 2. This update allows the tool and library to work with a wider range of models and content types. The support for embeddings is an important feature of the tool and library.
Simon Willison's project has been affected by a flood of AI-assisted pull requests, but he has not found this to be a problem. In fact, he has found the AI-assisted pull requests to be helpful, using tools like Opus 5.5 or GPT-6 Astra to triage and spot the most useful fixes. This approach has allowed him to efficiently review and implement the fixes.
The new release includes fixes from 12 authors, almost all of whom were assisted by AI. Simon Willison has expressed his gratitude to these authors for their contributions to the release. The fixes and updates in the new release are the result of a collaborative effort between Simon Willison and the contributing authors.
The release of the new version of the LLM CLI tool and Python library is a significant update that provides users with new features and improvements. The update is expected to make the tool and library more useful and accessible to users. Simon Willison's approach to managing AI-assisted pull requests has allowed him to efficiently review and implement fixes, and the new release is a testament to the success of this approach.
Questions and answers
What is the new default model in the LLM CLI tool and Python library?
The new default model is gpt-6-luna, which is half the price of the previous default model, gpt-5.6-luna.
How has Simon Willison's project been affected by AI-assisted pull requests?
Simon Willison's project has been affected by a flood of AI-assisted pull requests, but he has not found this to be a problem and has actually found it to be helpful.
How many authors contributed to the new release of the LLM CLI tool and Python library?
The new release includes fixes from 12 authors, almost all of whom were assisted by AI.
