- Published on
Llm-gemini 0.35
- Authors
- Name
- DeepHorus
The llm-gemini plugin released version 0.35 on 13th October 2026, adding support for Gemini 4 Argon. The model became available the same day via the Gemini API and to subscribers of a $100 per month plan. The update gives command-line users immediate access to Google's newest model.
Gemini 4 Argon was tested with the plugin's standard SVG pelican prompt at low, medium and high effort settings. All three illustrations were judged excellent, consistent with Gemini's strong record on SVG generation. The high-effort pelican cost 17.166 cents to produce.
Launch pricing is $2 per million input tokens and $10 per million output tokens. That matches GPT-6 Sol, halves Claude Opus 5.5's price, and is one fifth of Claude Fable 5.1 and GPT-6 Astra. On 1st December, Argon's price doubles, matching Opus 5.5.
Per-token prices are hard to compare, because what matters is how many tokens a model consumes to finish a task inside an agentic harness. Artificial Analysis publishes the total cost of running its benchmark with each model, and Argon costs $2,407 — close to GLM-5.3 and well below the latest OpenAI, Anthropic and Meta releases. After December's increase it should still undercut GPT-6 Astra.
Google's DeepMind staff have invested specifically in SVG work and have highlighted pelican outputs before. Bonus examples include an SVG armadillo in fishnet tights jaywalking on Mars and a Fabergé egg modelled in Blender. Staffer @fofrAI also shared a pelican riding a bicycle built from Fourier epicycles.
The launch reflects intensifying competition among frontier labs over both price and token efficiency. Buyers weighing Argon against rivals must now account for a December price change and differing agentic token consumption. Argon's benchmark economics look competitive for now, but that position depends on how workloads evolve.
