llm-gemini 0.34

Version 0.34 of llm‑gemini adds Google’s newly announced model, Gemini 3.8 Flash, which is available in three reasoning tiers: low, medium, and high. The update also fixes a bug that prevented the model version from being recorded in asynchronous responses, thanks to the work of Charlie Tonneslan. Gemini 3.8 Flash, along with its “Flash Cyber” variant reserved for trusted users, stands out for its speed and low cost, making it useful for tasks such as generating HTML and JavaScript code. In a test, the author asked for “a cool element in HTML” and the model delivered a working result in 13 seconds, at an estimated cost of about 1.8 cents.

The announcement includes visual comparisons between the pelicans generated by Gemini 3.8 Flash and the previous 3.7 Flash, showing improvements in output quality. The author also integrated the new model into his markdown‑svg‑renderer tool, which can now process HTML blocks within Markdown via an isolated iframe, expanding its interactive content rendering capabilities.

This news is significant because a fast, inexpensive model broadens the possibilities for automating web‑development and content‑generation tasks, enabling developers and creators to incorporate AI into everyday workflows. In the digital‑market context, tools like these can accelerate interface and catalog creation, benefiting platforms such as dataqbs’ open‑garage.

Read the original article on Simon Willison

This summary is an informational synthesis produced by dataqbs.com. All rights to the original content belong to its author and the cited media outlet. We act solely as curators of technology news and claim no authorship.