Gemini 4 Argon

The model was announced September 30 and first went only to trusted cyber defenders in the Fairwind Program. General availability followed on October 13 through the paid Gemini API, AI Studio and Google Cloud, with Ultra subscribers also getting access in Antigravity. Launch pricing is $2 per million input tokens and $10 per million output tokens. A third-party writeup says that rises to $4 and $20. Output tokens are capped at 1 million, up from 64K. Artificial Analysis rates it level with OpenAI's GPT-6 Astra on its Intelligence Index at about 60 percent of the cost per task, though one analysis places it behind Claude Opus 5.5. A companion robotics post reports 87 percent on the ERQA embodied-reasoning benchmark. Google's demo had the model watch a video of cornhole, write MuJoCo simulation code, and transfer the result to Apptronik's Apollo humanoid, which landed seven straight throws. Google says this cuts weeks of simulation work to hours. The scores are company-reported, and the demo covers one task. For founders, the price gap against Astra, which costs $10 and $50, and the long output window make a case for re-running coding and agent workloads on Argon. The introductory pricing and unverified benchmarks argue for testing on your own tasks first.