Gemini 4 Argon: price, context window and specs

What Google says it's for

Google's model page calls Argon its flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, autonomous multi-step agents and long decode of up to 1M tokens. In its 30 September announcement, Google says it raised the output limit to 1M tokens, up from 64K, and released Argon first to trusted cyber defenders through its Fairwind Program. Broad availability followed on 13 October 2026.

Where you can use it

• Gemini app: Google AI Pro and Ultra subscribers, since 13 October 2026, per Google's Gemini app post.
• Gemini API: paid tier only, in Google AI Studio and Android Studio, per the developer post.
• Google Antigravity, Google's agent-first development platform: Google AI Ultra subscribers, who can also try Argon in AI Studio.
• Gemini Enterprise: generally available since 13 October 2026 in the web app, mobile app and AI developer tools. In the app the toggle is off by default and defaults to on from 27 October 2026, per the release notes. In AI developer tools it's on by default for licensed users unless the organization uses a custom model list.

Pricing details

• Standard paid tier: $2 per million input tokens and $10 per million output tokens through 30 November 2026, then $4 / $20 from 1 December 2026, per Google's pricing page. Output prices include thinking tokens.
• Caching is implicit only; explicit caching isn't supported. Cached tokens cost $0.10 per million through 30 November, then $0.20.
• Batch and Flex: $1 / $5 through 30 November, then $2 / $10. Priority: $3.60 / $18, then $7.20 / $36.
• Google lists no separate price tier for long prompts.
• There's no free tier. The Contributor tier ($1 / $5 through 30 November, then $2 / $10) is an invitation-only private preview, and Google's table says content sent on it is used to improve its products.
• Grounding with Google Search: 5,000 free requests per month, shared across Gemini 3 and newer models, then $14 per 1,000.

API specs

• Model ID gemini-4-argon, a stable version.
• Context window: 1,048,576 tokens, shared by input and output. Default synchronous output is 65,536 tokens; long decode raises it to 1,048,576.
• Input: text, image, video, audio and PDF. Output: text only. Image generation, audio generation and the Live API aren't supported.
• Thinking levels: low, medium and high.
• Supported: function calling, structured outputs, code execution, file search, URL context, grounding with Google Search and Google Maps, and computer use (preview). Batch, Flex and Priority inference are available.

Google's benchmark claims

These are Google's own figures from its 30 September post, not independent results: 77.9% on DeepSWE v1.1 (long-horizon software engineering), 51.3% on Zapier's AutomationBench, 91.7% on LVBench (long video understanding) and 68% on CWE-bench v1, where Google says Argon ties for first place. Results from open datasets (Arena, Epoch AI) are on our benchmarks page.