Gemini 4 Argon Long Decode: 200K Tokens, One Request

Until this week, every Gemini call had a ceiling: about 64K tokens of output. If you wanted a full migration runbook, a book-length report, or a codebase-wide refactor plan, you chunked the work by hand and stitched it back together. On October 13, 2026 Google made Gemini 4 Argon broadly available, and the part developers are passing around is long decode: one request that can generate up to 1,048,576 tokens, with the server doing the stitching for you.

This guide shows how to use it from Python through the Interactions API: start a background run, poll it safely, extract the text, and (the part most posts skip) predict what it will cost before you press go. The cost model matters because every 64K segment of output re-bills everything that came before it. By the end you will have a small, tested module you can drop into a project.