Gemini 4 Argon
Gemini 4 Argon is Google's newest flagship reasoning model, generally available in the Gemini API since October 13, 2026. It reads text, images, video, audio and PDFs across a 1M-token window and writes text. Google positions it for long coding,…
Best for
Long, complex software engineering, agentic, math, science and enterprise knowledge-work tasks where top-tier reasoning matters more than cost.
Overview
Gemini 4 Argon is Google's newest flagship reasoning model, generally available in the Gemini API since October 13, 2026. It reads text, images, video, audio and PDFs across a 1M-token window and writes text. Google positions it for long coding, agent, math/science and enterprise tasks.
Strengths and limits
- Leads Google's own comparison table on most rows, incl. Vals Index and DeepSWE
- 1M-token context with very strong long-context retrieval (GraphWalks)
- Full Gemini API tool set: search grounding, code execution, function calling, computer use (preview)
- Text output only; no image/audio generation or Live API
- Introductory API price doubles on Dec 1, 2026; not on free tier
Published benchmark scores
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →