Gemini 3.5 Pro is the most anticipated unreleased AI model of 2026. Google’s CEO Sundar Pichai confirmed it at Google I/O on May 19 — then told the audience to “give us until next month,” drawing an audible groan from developers expecting the launch that day. We’re now in the final stretch of June with less than two weeks left in that window. Here’s everything confirmed, what’s still expected, and why this launch matters.
Current status: still not released
As of June 19, 2026, Gemini 3.5 Pro remains in limited preview for select Vertex AI enterprise customers only. It has not shipped to the public Gemini app, Google AI Studio, or the consumer Gemini subscription. Pichai’s exact words at I/O: “We’re also hard at work on 3.5 Pro. It’s already being used internally, and we look forward to rolling it out next month.”
That “next month” meant June 2026. With 11 days remaining, the window is closing. Polymarket prediction markets currently give roughly 50–55% odds for a pre-June 30 release.
Confirmed specs
| Feature | Gemini 3.5 Pro | Gemini 3.5 Flash (live now) |
|---|---|---|
| Context window | 2 million tokens | 1 million tokens |
| Reasoning mode | Deep Think | Standard |
| Multimodal | ✅ Full | ✅ Full |
| Expected pricing | ~$15/$60 per 1M tokens | $1.50/$9 per 1M tokens |
| Consumer access | $20/mo Gemini Pro plan | Free (default) |
| Deep Think access | $250/mo Ultra plan only | N/A |
Why the 2-million-token context window is a big deal
Two million tokens is the largest context window of any production frontier model announced to date. To put it in concrete terms:
- Roughly 1.5 million words — five to eight full novels
- An entire large codebase held in memory at once
- A full year of meeting transcripts or email threads
- ~30 minutes of video at standard sampling
- Hundreds of PDF documents processed simultaneously
For enterprise use cases — legal document review, codebase-wide refactoring, long-horizon research — this is the specification that changes what’s possible in a single session.
What “Deep Think” reasoning mode means
Deep Think is Google’s name for extended reasoning — the model spends more time evaluating a problem before generating an answer, similar to OpenAI’s o3 reasoning mode. The practical effect: significantly better performance on hard mathematical reasoning, complex multi-step logic, scientific problem solving, and ARC-AGI-2 style tasks.
The catch: Deep Think is restricted to the $250/month Ultra plan. Standard Gemini Pro subscribers at $20/month get access to the 2M context window and full model capability, but not the extended reasoning mode.
How it compares to Claude Opus 4.8 and GPT-5.5
| Gemini 3.5 Pro | Claude Opus 4.8 | GPT-5.5 | |
|---|---|---|---|
| Context window | 2M tokens | 200K tokens | 1M tokens |
| Reasoning mode | Deep Think | Extended thinking | o1-style reasoning |
| Input pricing (est.) | ~$15/1M | $5/1M | ~$15/1M |
| Coding benchmark | Expected top 3 | 69.2% SWE-bench | ~70%+ SWE-bench |
| Multimodal | ✅ Full | ✅ Full | ✅ Full |
Gemini 3.5 Flash — the model Google shipped at I/O — already scored 76.2% on Terminal-Bench 2.1 and 83.6% on MCP Atlas, outperforming Gemini 3.1 Pro on coding and agentic benchmarks while being faster and cheaper. Pro is expected to close the remaining gaps on hard reasoning and long-context retrieval where Flash still trails.
Who should wait for Gemini 3.5 Pro vs use something now
Wait if you: need the 2M context window for large document or codebase work, care specifically about Google’s ecosystem (Antigravity, Vertex AI, Android Studio), or want to compare all frontier models before committing to an annual plan.
Don’t wait if you: primarily do coding or agentic tasks (Gemini 3.5 Flash already beats last-generation Pro on these at lower cost), work on tasks under 200K tokens where context length doesn’t matter, or need a production deployment decision today.
When will it actually launch?
Every previous Gemini model launched via a single blog post on blog.google with a full benchmark grid — no staged rollouts or Twitter teasers. Watch for that post. If you want push notification the moment it ships, subscribe to the Gemini API changelog at ai.google.dev. The model ID will resolve in Google AI Studio simultaneously with the announcement.
Based on prior patterns and Pichai’s June commitment, the most likely window is the final two weeks of June 2026. If it slips past June 30, expect a formal timeline update from Google DeepMind.
This page will be updated the day Gemini 3.5 Pro launches with full benchmark results and a complete review.
Related reading
- Cursor AI review — already supports Gemini 3.x via the model picker; will add 3.5 Pro on availability
- GPT-5.6 benchmark leaks — the other June 2026 model release everyone is watching
- GPT-5.6 full guide — how it stacks up as context window and reasoning benchmark competition heats up