Google promised Gemini 3.5 Pro would reach general availability in June 2026. It didn’t. As of June 27, the model sits in limited Vertex AI enterprise preview, and Google has pushed the public launch to July. That delay matters, but it’s not the lead story. Four senior Gemini researchers announced they are joining Anthropic within the past six days. A missed deadline and a talent wave leaving simultaneously is a pattern that developers building on Google’s AI stack need to understand.
What Google Actually Promised and When
At Google I/O on May 19, 2026, Sundar Pichai introduced Gemini 3.5 Pro to a packed audience. When asked about general availability, Pichai said, “give us until next month.” The audience audibly groaned. That reaction tells you something: expectations had already been tempered by Google’s track record in 2026.
This is Google’s second major AI delivery miss this year. Gemini Ultra 1.5 was delayed by three months in 2026. That model eventually shipped and performed well, but the pattern of promising dates and missing them is now established. June came and went. The public API remains closed. Enterprise preview customers on Vertex AI can apply for early access, but the general developer community is waiting on a July date that carries no guarantee.
The Confirmed Specs: Why Developers Were Waiting
The anticipation is justified by what Gemini 3.5 Pro is confirmed to offer. These specs distinguish it from everything currently available at general access:
- 2,000,000 token context window. This doubles Claude Opus 4.8 and most competitors. For developers building on long-document analysis, codebase-level reasoning, or multi-session agents, this is a genuine capability gap, not a benchmark number.
- Deep Think reasoning mode. Google’s equivalent of extended thinking, similar to OpenAI’s approach. This enables multi-step reasoning on hard problems without prompt engineering workarounds.
- Frontier multimodal capabilities. Text and image inputs are confirmed. Additional modalities are expected at launch.
- Distribution plan. Google AI Studio, Vertex AI API, and direct API access are all on the roadmap for general availability.
- Pricing estimate: approximately $15 input / $60 output per 1M tokens. That is roughly 10x the cost of Gemini 3.5 Flash. It positions the model squarely for high-value, long-context workloads, not general inference.
For comparison, Gemini 3.5 Flash is already generally available. It beats Gemini 3.1 Pro on coding and agentic benchmarks at approximately four times the speed. Its pricing stands at $1.50 input / $9.00 output per 1M tokens, though that rate tripled from the prior Flash tier. Flash is the right choice for most developers right now. Pro is for specific use cases where the 2M context window or Deep Think reasoning changes what’s possible.
Four Researchers Out the Door: The Talent Story
The delay is a scheduling problem. The researcher’s departures are a different kind of signal.
In the week of June 21-27, 2026, four senior Gemini researchers announced they are leaving Google for Anthropic. This follows a broader pattern across 2025 and 2026, where Google has lost key AI researchers to Anthropic, OpenAI, and startups. The timing here is specific: these announcements came while Gemini 3.5 Pro was still in preview and the June GA target had just been missed.
Senior researchers leaving before a major model ships is a meaningful signal. The reasons could be disagreement on technical direction, frustration with execution velocity, or simply better opportunities elsewhere. Anthropic’s research culture and compensation have attracted significant talent in the past two years. Whatever the individual reasons, the aggregate effect is clear: Anthropic gains researchers who worked directly on Gemini’s architecture. Those researchers will influence Claude’s next generation.
This does not mean Gemini 3.5 Pro is weak. Google has a large and deep research organization. Individual departures do not hollow out a team of that scale. But the confluence of a missed public deadline and a visible talent wave in the same week is nothing. Developers evaluating long-term platform bets should log it as a data point, not dismiss it.
Gemini 3.5 Pro vs. the Current Field
| Model | Context Window | Pricing (input / output per 1M tokens) | Status |
|---|---|---|---|
| Gemini 3.5 Pro | 2M tokens | ~$15 / $60 (est.) | Limited enterprise preview |
| Claude Opus 4.8 | 1M tokens | $5 / $25 | Available (Fable 5 offline) |
| GPT-5.5 | 1M tokens | $5 / $30 | Available |
| Gemini 3.5 Flash | 1M tokens | $1.50 / $9 | Generally available |
| GPT-5.6 Sol | Unknown | Unknown | Government-gated (20 partners) |
The context window advantage is Gemini 3.5 Pro’s primary differentiator at this moment. Claude Opus 4.8 and GPT-5.5 are both available today at roughly one-third the output price. If your use case fits within 1M tokens, there is no reason to wait for Gemini 3.5 Pro. If you need the 2M window, there is currently no alternative, which means you either wait or restructure your architecture.
What Developers Should Do Right Now
- Use Gemini 3.5 Flash as your interim model. It is generally available, faster than Gemini 3.1 Pro, and cheaper per token for most workloads. If your application does not require the 2M context window or Deep Think reasoning, Flash handles the majority of coding and agentic tasks well.
- Apply for Vertex AI enterprise preview access now. If you have a genuine long-context use case, get on the waitlist. Google Cloud enterprise preview customers can access Gemini 3.5 Pro before general availability. The earlier you apply, the more lead time you have for evaluation.
- Pressure-test your context window requirements. The 10x pricing premium over Flash is significant. At $60 per 1M output tokens, a production workload that burns 10M output tokens per day costs $600 daily. Before committing to Gemini 3.5 Pro, confirm that the 2M window is genuinely necessary and that the unit economics work at that price point.
- Do not rebuild your architecture around a July date. Google has missed two major delivery targets this year. July 2026 is the current guidance, not a commitment. Design your roadmap to treat GA as a bonus event, not a dependency.
- Watch the researcher news closely. If additional senior departures follow in the next two to four weeks, that changes the calculus on long-term platform trust. Four departures in one week is notable. Ten departures over a month is a different conversation.
The Bottom Line
Gemini 3.5 Pro slipped to July, and four senior Gemini researchers just joined Anthropic. Those two facts together deserve a direct read: Google has a real execution problem in 2026, and the talent moving to Anthropic is building the models that will compete against it. The model itself is likely strong. Google’s history shows that when it eventually ships, the technical quality is usually there. Gemini Ultra 1.5 proved that. But the 2M context window advantage narrows every month that general availability slips, as competitors continue shipping. Developers with long-context use cases should apply for Vertex AI preview access and build for July without depending on it. Everyone else should build on Gemini 3.5 Flash today and evaluate Pro when it actually ships.
