Google reportedly delays Gemini 3.5 Pro to July to keep tuning the model

Google has reportedly delayed Gemini 3.5 Pro to July. The launch was expected in June, so the slip is roughly a month. According to a report from Business Insider, Google is taking the extra weeks to keep tuning its next frontier model before a wider release.

The slip is notable because Google already shipped the smaller, faster member of the same family. Gemini 3.5 Flash launched at Google I/O in May, so the company is sitting on a released Flash tier while the flagship Pro tier keeps cooking. That gap, between a model that’s out and a model that isn’t, is the whole story here, and it lands right as OpenAI and Anthropic push their own mid-year releases. Google is choosing a later, better-tuned frontier model over a faster headline.

What we know

The reporting is consistent across several outlets, even if Google won’t confirm any of it on the record yet.

• The delay is from June to July, per Business Insider’s source, described as a person familiar with the matter.
• Google is reportedly using the extra time to gather input from early testers on its Antigravity platform and the LMArena benchmarking site, Crypto Briefing reported, then tuning the model on that feedback.
• Token efficiency is reportedly part of it. Flash users complained the model burns through tokens too quickly, raising costs, and Google is said to be folding those lessons into Pro before it ships.
• Gemini 3.5 Flash is already public. It arrived at I/O in May as the first model in the 3.5 family, which is why the Pro delay reads as a tuning call rather than a stumble across the whole line.

Google previewed the new model at I/O, where CEO Sundar Pichai said it would arrive the following month. “Next month,” he reportedly said on stage, while the company noted at the time that the model wasn’t ready. That caveat now looks load-bearing.

What we don’t know

Plenty. Business Insider’s June 24 report leans on one anonymous source, so four soft spots are worth naming.

• The exact July date. “July” is a month, not a ship date. Nobody has a firm day.
• The real reason. Tuning is the stated cause, but the timing also overlaps a reported talent exodus, with senior researchers said to be leaving for OpenAI and Anthropic. How much that bleeds into the schedule is unclear.
• The benchmark targets. We don’t know what scores Google is chasing before it’s comfortable shipping Pro, or which evals it failed in early testing.
• Whether July holds. A model that’s already slipped once can slip again. There’s no guarantee the next date is the last one.

Who reported this

Business Insider broke the delay, citing a single source familiar with the plans, and the report moved through Reuters’ wire onto financial-news sites. Crypto Briefing and Analytics Insight corroborated the timing and the stated reasons. Google itself stayed quiet. A company spokesperson declined to comment on the revised schedule, which is why every claim here is hedged. Treat the July target as a reported plan, not a promise.

What this means for you

If you’re building on Gemini, don’t wire a launch around “Pro in July” yet. Flash is the model you can actually ship against today, and it’s the one Google is iterating on in public. For anyone benchmarking across vendors, the timing is the real signal: Google is choosing a later, more-tuned frontier model over a faster headline, which is a different bet than racing a half-baked flagship out the door. Compare that with how aggressively Anthropic pushed Claude Opus 4.8 out the gate and how fast DeepSeek shipped V4. My read: a one-month tuning slip is cheap insurance against a bad launch, and it tells you Google would rather lose a news cycle than ship a Pro tier that fumbles long, agentic tasks. Watch whether July holds. If it slips again, the tuning story stops being reassuring.