llm-gemini 0.35 Adds Gemini 4 Argon Support

The latest release of llm-gemini 0.35 introduces support for the new Gemini 4 Argon model, now available through the Gemini API and for users on the $100/month subscription plan.

AI-generated editorial illustration of the article topic; not a photograph, product screenshot or measured result.

Release of llm-gemini 0.35 with Gemini 4 Argon Support

The latest release of llm-gemini 0.35 introduces support for the new Gemini 4 Argon model, now available through the Gemini API and for users on the $100/month subscription plan. This update expands the capabilities of the llm-gemini plugin, which serves as an interface to Google's Gemini family of models. According to the release notes, the addition of Gemini 4 Argon is part of Google's ongoing efforts to enhance its AI offerings. The pricing for model usage, as noted in the release, includes a specific example where an illustration cost 17.166 cents. This information is attributed to the source publisher and should not be generalized beyond the context provided.

Pricing Details for Gemini 4 Argon

The pricing for Gemini 4 Argon starts at $2 per million input tokens and $10 per million output tokens at launch, matching the rates of GPT-6 Sol, according to Simon Willison. However, on December 1st, the price is set to double, aligning it with the cost of Claude Opus 5.5, as noted in Simon Willison. While these figures provide a baseline for comparison, Simon Willison highlights that modern evaluations of large language models should consider factors beyond raw pricing, such as the efficiency of token usage within an agentic system. This shift in focus underscores the importance of performance metrics in assessing value, rather than relying solely on cost per token.

Benchmark Cost Analysis for Gemini 4 Argon

Artificial Analysis provides insight into the total cost of running benchmarks with various models, revealing that Gemini 4 Argon costs $2,407 to execute the benchmark. This places it near GLM-5.3 and far below the costs associated with models from OpenAI, Anthropic, and Meta’s latest offerings. According to the data, even if the price were to double, Argon would still remain slightly more affordable than GPT-6 Astra. These figures highlight the economic advantages of Gemini 4 Argon in benchmark execution, as reported by Artificial Analysis.

Example Outputs and Performance of Gemini 4 Argon

The Gemini 4 Argon model has demonstrated versatility in generating complex and creative outputs. Examples include an SVG of an armadillo in fishnet tights jaywalking on Mars and a Fabergé egg created using Blender. A DeepMind staffer, @fofrAI, shared a video of a pelican riding a bike using Fourier epicycles, showcasing the model's ability to handle intricate visual and mathematical tasks. These examples highlight the model's potential in creative and technical applications. However, specific performance metrics or benchmarks have not been provided by the source, and claims about its capabilities should be attributed to the publisher or user who shared them.