July 17 — today — has been the widely reported target for Gemini 3.5 Pro’s general availability. But a Bloomberg report picked up by 9to5Google last night (July 16) says the date may slip again — Google is reportedly still working through coding-performance issues, and no confirmation has landed as of this writing. Given the pattern so far (June target → July target → now a July 17 target in doubt), treat today’s date the same way we’ve been treating every other Gemini 3.5 Pro claim: unconfirmed until Google actually publishes a model card.
What Bloomberg reported
According to Bloomberg, Google is “taking time to try to improve [Gemini 3.5 Pro’s] capabilities, particularly in coding.” In late June, Google reportedly updated the training data specifically to address coding weaknesses, but internal results were described as disappointing. That timeline points to a real reset between the original May 19 Google I/O announcement and today — not just a scheduling slip, but a substantive capability gap the team hasn’t fully closed.
The pattern so far
Sundar Pichai announced the Gemini 3.5 family at I/O on May 19, shipping only Gemini 3.5 Flash that day and telling developers to “give us until next month” for Pro — a line that reportedly drew an audible groan from the audience. June came and went with no Pro launch. Reports then said Google had scrapped the original 3.5 Pro base model entirely and restarted pretraining on a native Gemini 3 foundation, pushing the target to July 17 — the same day as Shanghai’s World AI Conference. Now that date itself is in question.
Why it matters competitively
Every week Gemini 3.5 Pro slips, the field around it gets more crowded. GPT-5.6 Sol reached GA on July 9. Grok 4.5 opened to the public the same day. DeepSeek’s V4 family targets its own stable release around the same window. If Gemini 3.5 Pro slips past today, Google will be launching its flagship reasoning model into a market where three other frontier options have already had a running start — a materially different competitive position than if it had shipped on the original June target.
What’s actually confirmed right now
As of this update, the public Gemini API still lists gemini-3.5-flash and gemini-3.1-pro-preview — no gemini-3.5-pro model ID, no model card, no pricing page. Gemini 3.5 Flash remains the model actually carrying production workloads, with confirmed benchmarks (76.2% on Terminal-Bench 2.1) and pricing ($1.50 input / $9 output per million tokens). Everything about Pro — the 2M-token context window, Deep Think gating, benchmark numbers — is still third-party reporting, not an official Google claim.
Last updated: July 17, 2026. We’ll update this post the moment Google confirms a date or ships the model.