Gemini 3.5 Release Date: Flash Shipped, Pro Never Did

If you searched for “Gemini 3.5 release date” or “what’s new in Gemini 3.5,” the honest answer has two halves. The Flash half shipped — repeatedly. The Pro half didn’t, and as of October 5, 2026 Google still hasn’t given it a date. This page is the status/timeline record for the whole Gemini 3.5 family: exactly what launched, exactly what got delayed, and what you should actually be calling in the API today. Figures below are re-verified against Google's model list and pricing page on 5 October 2026, and every Artificial Analysis score carries its index version — Artificial Analysis rebased its Intelligence Index from v4.1.1 to v4.3.2 in October 2026 and published no conversion between them, so older scores quoted elsewhere will not match these.

Is Gemini 3.5 released yet?

Partially. Here is the family status as of October 5, 2026:

• Gemini 3.5 Flash — released. GA on May 19, 2026, day one across the Gemini app, Search AI Mode, Antigravity, the Gemini API, AI Studio, Android Studio, and Gemini Enterprise. API model ID gemini-3.5-flash. Pricing: $1.50 input / $9.00 output per 1M tokens.
• Gemini 3.5 Flash-Lite — released. July 21, 2026, at $0.30/$2.50 per 1M tokens, running around 350 tokens/second.
• Gemini 3.5 Flash Cyber — limited access. Announced July 21, 2026: a cybersecurity fine-tune of 3.5 Flash inside the CodeMender agent framework, available only as a pilot for governments and trusted partners.
• Gemini 3.5 Pro — NOT released. Announced at I/O on May 19 with a “next month” promise, then delayed past June, then delayed indefinitely. There is still no gemini-3.5-pro model ID anywhere, no model card, and no pricing.

And the family didn’t stop at 3.5: Gemini 3.6 Flash shipped July 21, Gemini 3.7 Flash on August 13 and Gemini 3.8 Flash on September 2 — all three at an introductory $0.75/$3.75 per 1M tokens. Google’s big-model story through 2026 is a roughly monthly Flash cadence, while the Pro tier you can actually call is still Gemini 3.1 Pro (preview) from February. Four Flash releases, zero Pro releases, in a little over three months.

What is the full Gemini 3.5 release timeline?

Nothing else in the 3.5 line has shipped in this window, and no Gemini model — 3.5 or otherwise — currently offers the 2M-token context that early 3.5 Pro speculation promised. Gemini 3.1 Pro and 3.7 Flash are both 1M-token models.

What’s new in the Gemini 3.5 Flash line?

Gemini 3.5 Flash was a genuine frontier-grade Flash release: 1M-token context, full multimodal input, and day-one availability across every Google surface. The catch was the price. At $1.50/$9.00 per 1M tokens it landed roughly 3x higher than its Flash predecessor, and the developer community read it as mispriced.

Google’s answer came fast. Gemini 3.6 Flash (July 21) improved coding benchmarks (DeepSWE 49% vs 37%, Google self-reported) while producing 17% fewer output tokens — and launched with “introductory pricing” at $0.75/$3.75 through December 31, 2026, half of 3.5 Flash’s rate. Gemini 3.7 Flash (August 13) kept that intro price and pushed further: 39.06 on Artificial Analysis's Intelligence Index v4.3.2, up from Gemini 3.6 Flash's 33.98 — a 5.08-point jump, by far the largest step in the Flash line — and, on Google's own published benchmarks, scores that beat its Pro line. Gemini 3.8 Flash (September 2) added only 1.87 points on top, to 40.93, and gave up throughput and first-token latency to do it. The practitioner read holds: in this family, Flash is better than Pro, and the big gain was 3.6 to 3.7 rather than 3.7 to 3.8.

For scale on the independent numbers, here is the whole callable Flash line on one index version:

Read the Pro row carefully, because it is the clearest evidence for the whole "Flash beats Pro" claim: Gemini 3.1 Pro Preview scores 29.72, below every Flash model in the 3.5-and-later line except Flash-Lite — while costing $2/$12 per 1M tokens against Flash's $0.75/$3.75. Google's Pro tier is currently both more expensive and less capable than its Flash tier on independent measurement. That is the real story of Gemini 3.5, and it is why the missing 3.5 Pro mattered.

The net effect is awkward for 3.5 Flash itself: it is now the most expensive and least capable of the current Flash trio. If you adopted it in May, moving to 3.7 Flash is a price cut and an upgrade in one change. Full specs, benchmarks, and migration notes in our Gemini 3.7 Flash launch guide.

Note the intro-pricing fine print: 3.6, 3.7 and 3.8 Flash all revert to $1.50/$7.50 per 1M tokens on January 1, 2027. Every discount tier doubles with it — batch and flex to $0.75/$3.75, priority to $2.70/$13.50, cached input reads to $0.15 per 1M, and cache storage to $1.00 per 1M tokens per hour. Budget against the standard rate, not the promo.

What happened to Gemini 3.5 Pro?

It missed its window, publicly. The promise trail:

1. May 19: announced at I/O — in internal use, “rolling it out next month.”
2. Late June: the June target passes; Business Insider reports a slip to July.
3. July 16: Bloomberg and CNBC report the delay is about coding performance — late-June training runs with updated data yielded disappointing results. Alphabet’s stock dipped on the news.
4. July 21: Google’s official line softens to “as soon as it’s ready,” with no date.
5. August 13: Google ships 3.7 Flash instead, with no mention of Pro.

As of October 5, 2026: no API model ID, no model card, no pricing, no date. Google's public model list has moved on to Gemini 3.8 Flash without ever filling the 3.5 Pro slot. The Pro tier you can actually call remains Gemini 3.1 Pro (preview) at $2/$12 per 1M tokens (up to 200K context; $4/$18 above that) — and it scores 29.72 on Artificial Analysis's Intelligence Index v4.3.2, below every current Flash model. For the full delay analysis and what to use in Pro’s place, see our Gemini 3.5 Pro status page.

Which Gemini model should you use today?

The current, real, callable lineup as of October 2026:

The one-line recommendation: Gemini 3.8 Flash is the highest-scoring Gemini you can call, but Gemini 3.7 Flash is the better default — identical list price, 33% cheaper per completed task, faster to first token and faster to stream. Move to 3.8 Flash specifically for long-horizon asynchronous agentic work, where the extra latency costs you nothing. The Pro tier is still Gemini 3.1 Pro, which trails every current Flash model on independent benchmarks, and Gemini 3.5 Pro remains unreleased. Keep your model layer swappable behind config regardless — this family has changed its answer four times in five months. One footnote for completeness: a model called Gemini 4 Argon now appears on third-party leaderboards (Artificial Analysis scores it 52.56 on Intelligence Index v4.3.2, and it tops LMArena's text board at 1525 ±9, preliminary), but it is not in Google's public Gemini API model list as of 5 October 2026, so treat it as not generally callable yet.

FAQ

Is Gemini 3.5 Pro available in the API?

No. As of October 5, 2026 there is no gemini-3.5-pro model ID in Google's published Gemini API model list, no model card, and no announced pricing — almost five months after it was promised for “next month.” The Pro-tier model you can call is Gemini 3.1 Pro (preview) at $2/$12 per 1M tokens, which scores 29.72 on Artificial Analysis's Intelligence Index v4.3.2 against Gemini 3.8 Flash's 40.93.

When will Gemini 3.5 Pro be released?

Unknown. Google’s last official statement (July 21, 2026) was that it is “currently testing with partners” and will ship “as soon as it’s ready.” Bloomberg reported the delay stems from coding performance. Anyone citing a specific date is speculating.

Is Gemini 3.5 Flash still worth using?

Only if you’re already on it and can’t migrate yet. At $1.50/$9 it costs double what 3.7 and 3.8 Flash charge during the intro period ($0.75/$3.75), and it scores 32.60 on Artificial Analysis’s Intelligence Index v4.3.2 against 39.06 for 3.7 Flash — while costing more per completed index task ($1.56 against $0.93). It is the worst value in the family on every axis. Migration is a model-ID change.

What is Gemini Spark?

A persistent personal agent built on the Gemini Flash line, announced at I/O 2026 — it “runs 24/7” taking actions on your behalf. It started with trusted testers and a US-only Ultra beta; as of August 13, 2026 it’s available to Google AI Pro and Ultra subscribers in 160+ countries.

Does any Gemini model offer a 2M-token context window?

No. The 2M-token figure was pre-launch expectation for Gemini 3.5 Pro, which never shipped. Gemini 3.7 Flash, Gemini 3.8 Flash and Gemini 3.1 Pro all top out at 1,048,576 input tokens — one megatoken — with a 65,536-token output ceiling on the Flash models.

What are Gemini 3.6 Flash, 3.7 Flash and 3.8 Flash?

The three Flash releases that followed 3.5 Flash: 3.6 Flash on July 21, 2026 (better coding, 17% fewer output tokens, 33.98 on AA Intelligence Index v4.3.2), 3.7 Flash on August 13, 2026 (39.06 — the biggest single jump in the line), and 3.8 Flash on September 2, 2026 (40.93, but slower on both latency axes and 33% dearer per completed task). All three launched at an introductory $0.75/$3.75 per 1M tokens through December 31, 2026, reverting to $1.50/$7.50 after.

Hiring developers who can move with the model field?

Four Flash releases and one indefinitely delayed flagship in five months is exactly why your model layer should be swappable — and why your team should evaluate releases on evidence rather than keynote promises. codersera.com/hire matches you with vetted remote developers experienced with LLM integration, evaluation, and migration, so adopting the right model stays a config change instead of a rewrite.