# Gemini 4 Argon costs $2 per million input tokens until November 30. Budget your agents at December's $4.

*2026-10-18 ⬩ 2:30 PM ET | bryn*

Google opened Gemini 4 Argon to developers on October 13 at $2 per million input tokens and $10 per million output, and the price doubles on December 1, so I'd budget agent runs at the December rate today. [Google's Gemini Developer API pricing page](https://ai.google.dev/gemini-api/docs/pricing) lists $4 input and $20 output starting December 1, 2026. Output includes thinking tokens. That matters most for agents, because a long run that reasons a lot pays for its thinking at the output rate.

[Simon Willison's October 13 post on his weblog](https://simonwillison.net/2026/Oct/13/llm-gemini/) makes the sharper point: what really matters is how many tokens a model takes to achieve a goal within an agentic harness.

Here's what I'd do. Run your real task end to end at the launch price, a code migration or a research pass. Log tokens per finished task, not per call. Multiply by the December rate. If the number still works, build on it. If it only works at $2, you've found a promotion, not a price.

There's no free tier for Argon, so set a budget cap before the first run. Cheap tokens spend fast.

## Sources

1. [Gemini Developer API pricing](https://ai.google.dev/gemini-api/docs/pricing). Google AI for Developers, read Oct 18, 2026.
2. [Release: llm-gemini 0.35](https://simonwillison.net/2026/Oct/13/llm-gemini/). Simon Willison's Weblog, Oct 13, 2026.

See also: https://www.civic.com/field-notes/ai-pricing-broke-this-year

Source: https://www.civic.com/bin/argon-price-doubles-december