Google announced a controlled preview of Gemini 4 Argon, initially limited to trusted cyber defenders in the Fairwind Program. It raises the single-trajectory output limit from 64K to 1M tokens. Introductory pricing is $2 per million input tokens and $10 output, scheduled to double later. Google reports leading coding, workflow, long-video and cybersecurity results, but the developer API is not yet available.
Google opens Gemini 4 Argon to paid developers
2026-10-01Verified announcementTRIAL
Google has opened Gemini 4 Argon to paid-tier developers through AI Studio and the Gemini API, with access also reaching Antigravity, Android Studio, the Gemini app and Search. The model supports a unified one-million-token input-output budget. Introductory pricing remains $2 per million input tokens and $10 output, doubling after November 30, 2026.
Official · Referenceblog.google · gemini 4 argonOfficial · AnnouncementGemini 4 Argon is now available
Availability: paid-tier developers in AI Studio and the Gemini API; Ultra or Pro access also varies across Antigravity, Android Studio, Gemini app and Search
What this means for you
For long-horizon agent teams, Argon has moved from a controlled preview to a callable model, making real task success, drift and cost measurable. The one-million-token limit is a shared input-output budget and pricing doubles on a fixed date, so trials should track token mix, human review and failure recovery together.
Evidence & related material
Google confirms paid-tier Gemini API access for Gemini 4 Argon and specifies the unified one-million-token budget plus pricing after November 30.
Subsequent changes
Gemini 4 Argon moves from controlled preview to the paid Gemini API tier and sets a date for its price increase.
Technical details & scores
| Availability | paid-tier developers in AI Studio and the Gemini API; Ultra or Pro access also varies across Antigravity, Android Studio, Gemini app and Search | ||||||
|---|---|---|---|---|---|---|---|
| Token Limit | 1M unified input-output tokens with a continuation protocol | ||||||
| Introductory Pricing Per Million |
| ||||||
| Pricing After 2026 11 30 Per Million | input 4 output 20 | ||||||
| Vendor Benchmarks | DeepSWE v1.1 77.9%; AutomationBench 51.3%; CWE-bench v1 68% | ||||||
| Controls | prompt-injection red teaming, chain-of-thought/action monitoring and hardened sandboxes | ||||||
| Time Basis | Google's official models feed published the availability update at 2026-10-13T16:00:00Z |
Technical impact 5
Engineering value 4
Adoption signal 3
Maturity 4
Verification cost 3
Risk 5