Google's Gemini 3.5 Pro is reportedly targeting a July 17 general-availability launch, according to leaked launch plans and third-party reporting rather than any official Google post. As of July 13, no model card, pricing page, or gemini-3.5-pro listing has appeared in the public API, so every headline number below should be read as reported, not confirmed.

What This Could Enable

The most builder-relevant rumor is a reported 2-million-token context window, double the 1-million cap on the current Gemini 2.5 Pro. If that ships, it would let you drop an entire mid-size codebase, a full shooting script, or a long research corpus into a single prompt without chunking, and keep a reasoning agent coherent across the whole thing. Reporting also points to a Deep Think reasoning mode for multi-step logic. Treat any workflow you design around these specs as provisional until Google publishes a model card, because the launch date itself is a leak.

Why It Matters for Creators

Context length is now a practical constraint on what creators can automate in one pass, from full-length video scripts to multi-file app builds. A verified 2-million-token frontier model would reset expectations for every competing tool, which is why the leak is worth tracking even before it is real, though the circulating specs still contradict each other. It also lands in a crowded July, alongside shipped releases like GPT-5.6 and Grok 4.5.

Key Details (Reported, Not Confirmed)

Target date: July 17, per TechTimes reporting that stresses every spec remains unconfirmed.

Context window: A reported 2 million tokens, up from 1 million.

Rebuild: Google reportedly scrapped its original base model and rebuilt from the ground up after engineers found failures in recursive tool-calling and SVG generation, as Startup Fortune noted.

Pricing: Leaks put input and output at roughly $1.25 and $10 per million tokens, though the figures vary by source.

What to Watch

Wait for an official Google DeepMind launch post and a live API listing before committing production work to Gemini 3.5 Pro. If the July 17 date holds, the first thing to test is whether the reported 2-million-token context stays coherent on real long-context tasks, not just whether it accepts the tokens.