Gemini 3.5 Pro Release Date: What Google Confirmed
The Direct Answer: What Is Confirmed Today?
| Question | Verified answer on August 12, 2026 | Evidence |
|---|---|---|
| Has Google named Gemini 3.5 Pro? | Yes | Google announcement and DeepMind family page |
| Is it released to the public? | No confirmed public release | DeepMind says “coming soon” |
| Is a public Gemini API route listed? | No | Gemini API model catalog |
| Is there a public model ID? | No | No endpoint in the model catalog |
| Is API pricing published? | No | No Gemini 3.5 Pro pricing row |
| Is there a new launch date? | No | Current official wording gives no date |
| Is an EvoLink route verified? | No | No authenticated request and route-verification record |
This is stronger evidence than a rumor round-up because it separates three states that search results often collapse: a named product, restricted partner testing, and a callable public API.
Where the Release-Date Claim Comes From
Google's May 19 Gemini 3.5 announcement said the team was using Gemini 3.5 Pro internally and looked forward to rolling it out “next month.” That created the June expectation repeated across news articles and search snippets. June passed without a public Pro release.
On July 21, Google's newer model update said Gemini 3.5 Pro was testing with partners and would become broadly available when ready. The DeepMind model-family page now uses the shorter “3.5 Pro coming soon” label. Neither statement supplies a revised calendar date.
The safe timeline is therefore:
| Date | Signal | What it proves | What it does not prove |
|---|---|---|---|
| May 19, 2026 | Google named Gemini 3.5 Pro and expected a next-month rollout | The model exists and was in development | A fixed GA date or API contract |
| July 21, 2026 | Google said partner testing was underway | Development progressed to restricted testing | Public Early Access or a callable endpoint |
| August 12, 2026 | DeepMind says “coming soon”; API records still omit Pro | The product remains pending publicly | Price, model ID, context, or launch day |
Third-party reports and community posts can explain why people search specific dates, but they cannot replace the current first-party status.
Gemini 3.5 Pro, Gemini 3.5 Flash, or Gemini 3.6 Flash?
The similar names create a real search problem. Gemini 3.5 Flash and Gemini 3.5 Flash-Lite are documented public models. Gemini 3.6 Flash is also a separate documented Flash model. None of those listings means Gemini 3.5 Pro has shipped.
Pro and Flash tier labels also do not let us copy specifications between models. A context limit, tool, price, or model code documented for Gemini 3.1 Pro or Gemini 3.5 Flash remains a baseline—not a Gemini 3.5 Pro fact.Release Does Not Mean API Availability
A Gemini release can appear through several surfaces: a Google announcement, the Gemini consumer app, partner testing, Google AI Studio, the Gemini Developer API, Vertex AI, or a third-party gateway. Those channels can move on different schedules.
For developers, a usable API release needs more than a product name. Verify all of the following:
- an official request model ID;
- a documented access channel and release stage;
- supported inputs, tools, context, and output limits;
- public pricing and quota terms;
- a successful authenticated request;
- returned model identity and usage metadata;
- billing that matches the published route.
gemini-3.5-pro string does not complete that chain.What Current Gemini Channels Teach Us
Current routes provide useful baselines while Pro is pending:
| Delivery need | Current model to evaluate | Why it helps now |
|---|---|---|
| Stable Gemini 3.5 API work | Gemini 3.5 Flash | Published endpoint, pricing, inputs, and measurable behavior |
| Current Pro-family work | Gemini 3.1 Pro | Documented Pro endpoint for complex reasoning and multimodal tasks |
| Lower-cost high-volume work | Gemini 3.5 Flash-Lite | A separate documented efficiency route |
| Family-level model selection | Gemini collection | Compare current routes without treating a pending model as inventory |
These models should be tested on your tasks. A product listing alone does not prove production readiness; verify request success, returned identity, usage, capacity, and billing for the route you plan to use.
What Developers Actually Want From the Next Pro Model
Search and community discussions focus on headline specifications, but production teams need harder outcomes:
- higher accepted-task quality on difficult coding, reasoning, research, and multimodal work;
- reliable long-horizon execution with fewer loops, invalid tool calls, and lost checkpoints;
- better token efficiency after retries, fallbacks, and human review are included;
- predictable structured output, tool schemas, quotas, safety blocks, and streaming;
- a migration path that does not force application logic onto one provider-specific model ID.
Claims about a 2M-token context window, Deep Think, exact pricing, or benchmark leadership remain unconfirmed for Gemini 3.5 Pro. They belong in a launch test plan, not a facts table.
Should You Wait? Choose by Delivery Risk
| Situation | Decision now | Why |
|---|---|---|
| A feature must ship this sprint | Use a verified current route | No public Pro API date exists |
| You need a current Pro baseline | Evaluate Gemini 3.1 Pro | It provides measurable behavior today |
| Your workload is latency- or volume-sensitive | Evaluate Flash-family routes | Pro may not be the best economic route even after launch |
| The future Pro model is strategically important | Save traces and define promotion gates | Preparation reduces launch-day integration risk |
| Procurement requires a fixed SKU and region | Wait for official commercial terms | Partner testing does not provide a public contract |
Waiting is sensible only when the work cannot proceed on any current model and a pending Pro capability is essential. For most teams, the safer strategy is to keep shipping and make the candidate replaceable through configuration.
The Seven Signals Worth Monitoring
The status changes only when evidence changes. Watch these seven signals:
- a Google launch post or DeepMind model card;
- a Gemini API catalog entry with an exact model code;
- a Gemini API release-note entry;
- a pricing row for standard and any supported service tiers;
- an AI Studio or Vertex AI access path with documented eligibility;
- published capability, quota, region, and deprecation terms;
- an EvoLink route that passes authenticated request, identity, usage, and billing verification.
The first six prove the upstream contract. The seventh proves the gateway route. Both layers matter for production access.
How EvoLink Handles Day-Zero Access
Before a live status change, EvoLink checks the upstream documentation and then validates a successful request, returned model identity, usage accounting, and billing. Teams can prepare now by keeping the candidate ID unset, freezing current baselines, and defining offline, shadow, canary, promotion, and rollback gates.
Sources
- Google DeepMind: Gemini model family
- Google: Gemini 3.5 announcement
- Google: July 21 model update confirming partner testing
- Google AI for Developers: Gemini API model catalog
- Google AI for Developers: Gemini API pricing
- Google AI for Developers: Gemini API release notes
FAQ
When is Gemini 3.5 Pro coming out?
Google has not published a current release date. The latest official status says the model is coming soon and testing with partners.
Has Google officially confirmed Gemini 3.5 Pro?
Yes. Google has named the model, and the DeepMind family page lists it as coming soon. That confirms the product, not a public API launch.
Is Gemini 3.5 Pro available in the Gemini API?
No public listing was present in Google's model catalog, pricing page, or API release notes when checked on August 12, 2026.
What is the Gemini 3.5 Pro model ID?
gemini-3.5-pro based on a URL, community label, or naming pattern.How much will the Gemini 3.5 Pro API cost?
Google has not published pricing. Current Gemini Pro and Flash prices are baselines, not forecasts for Gemini 3.5 Pro.
Will Gemini 3.5 Pro have a 2M context window or Deep Think?
Those fields are unconfirmed. Google has not published a Gemini 3.5 Pro model card or public API capability table.
Is partner testing the same as public Early Access?
No. There is no published public signup, eligibility policy, preview endpoint, or access guarantee for Gemini 3.5 Pro.
What should developers use while waiting?
Evaluate Gemini 3.5 Flash for a current stable Gemini 3.5 route or Gemini 3.1 Pro for a current Pro-family baseline, and keep model selection configurable.
