AI agents · OpenClaw · self-hosting · automation

Quick Answer

When Does Gemini 3.5 Pro Release? (June 2026 Target)

Published:

When Does Gemini 3.5 Pro Release? (June 2026 Target)

Gemini 3.5 Pro is targeted for release in June 2026, according to Google’s Gemini 3.5 announcement at Google I/O 2026 on May 19. Here’s everything known and rumored as of May 22, 2026.

Last verified: May 22, 2026

Quick facts

PropertyValue
StatusAnnounced at I/O 2026 (May 19, 2026)
Target releaseJune 2026
FamilyGemini 3.5 (alongside 3.5 Flash, shipped May 19)
Default surfaceGemini API, AI Studio, Gemini app
Expected context1,000,000+ tokens
Expected pricingUnannounced (likely $3–5 input / $15–30 output per 1M)
Deep Think modeYes, premium-only at launch
Free tier accessLikely later (free tier has 3.5 Flash now)

What Google actually announced at I/O 2026

Google’s Gemini 3.5 announcement on May 19, 2026 introduced the family, with two confirmed models:

ModelStatusDefault for
Gemini 3.5 FlashLive May 19, 2026Gemini app, AI Mode, Antigravity CLI
Gemini 3.5 ProRolling out next month (June 2026)Higher-quality reasoning, Deep Think

Mashable, AP, and Engadget all reported the June 2026 timing based on Google’s keynote messaging. Google did not provide a specific date.

Why Google split the family this way

Gemini 3.5 Flash hits an Artificial Analysis Intelligence Index around the mid-50s at one-third the price of Claude Opus 4.7, with frontier-class output speed. That’s the model Google wants powering Search (AI Mode), Antigravity CLI, the new Gemini app default, and information agents — anywhere scale matters more than peak reasoning.

Gemini 3.5 Pro is the answer to the other question: “When the task is hard, expensive, multi-step, or competition-grade, do you have the best model?” Pro is expected to:

  • Lead on competition coding benchmarks (HumanEval, SWE-Bench Verified)
  • Top math benchmarks (AIME, FrontierMath)
  • Handle multi-hour agentic loops better than Flash
  • Enable Deep Think — a Gemini 3 Pro reasoning mode that scaled up, accessible to Google AI Ultra subscribers
  • Push Google’s Artificial Analysis Intelligence Index into the high-60s, putting it competitive with GPT-5.5 and Claude Opus 4.7 at the top of the leaderboard

Pricing — what to expect

Google hasn’t disclosed Gemini 3.5 Pro API pricing as of May 22, 2026. Reasonable projection based on prior pricing patterns:

ModelInput / 1MOutput / 1MNotes
Gemini 3.5 Flash$1.50$9.00Confirmed May 19, 2026
Gemini 3.1 Pro$2.50$15.00Current Pro tier
Gemini 3.5 Pro (projected)$3–5$15–30Likely range based on Pro precedent
GPT-5.5$5.00$30.00OpenAI’s frontier
Claude Opus 4.7$15.00$75.00Anthropic’s frontier

Even at the top end, Gemini 3.5 Pro should remain meaningfully cheaper than Claude Opus 4.7 and roughly aligned with GPT-5.5, continuing Google’s pattern of competing on price-quality ratio.

Consumer pricing — Gemini app and Workspace

The new Google AI Ultra $100/mo tier announced at I/O 2026 is built for power users who’ll want 3.5 Pro:

TierCost / moGemini 3.5 FlashGemini 3.5 Pro (when live)Deep Think
Free$0Yes (low quota)Likely no initiallyNo
Google AI Pro$19.99YesYes (limited)No
Google AI Ultra $100$99.99Yes (priority)Yes (priority)Likely beta
Google AI Ultra $200$199.99Yes (priority)Yes (highest quota)Yes

Deep Think mode (the reasoning-extended variant) has historically been Ultra-exclusive at launch, then trickled to Pro.

How to get early access

There’s no public early-access program announced for 3.5 Pro as of May 22, 2026. Practical access path when it ships in June:

  1. AI Studio (ai.google.dev) — Developer playground, typically gets new models first. Free tier, rate-limited.
  2. Gemini API — Becomes available shortly after AI Studio.
  3. Gemini app — Premium subscribers (Ultra first, then Pro) get app access next.
  4. Antigravity 2.0 + Antigravity CLI — Will gain Pro as a selectable model.
  5. AI Mode in Search — Gets it last, gated to Personal Intelligence-enabled users in select regions initially.

Best signal to watch: Logan Kilpatrick’s Twitter, the Google AI dev blog, and the ai.google.dev/gemini-api/docs/changelog release notes.

How it will likely compare to GPT-5.5 and Claude Opus 4.7

Gemini 3.5 Pro (projected)GPT-5.5Claude Opus 4.7
AA Intelligence IndexHigh-60s (target)~60~62
Context window1M+1M200K
Default reasoning modeDeep Think (Ultra)ThinkingExtended thinking
SWE-Bench ProTBDStrongLeader
Cost (output / 1M)$15–30 (projected)$30$75
Best forBest price-quality ratioAgentic coding pipelinesHighest-stakes coding
Multimodal (video / image)Best in classStrongImproving

The pattern: Anthropic’s Opus 4.7 stays the gold standard for high-stakes single-shot coding; OpenAI’s GPT-5.5 stays the best for agentic pipelines; Google’s 3.5 Pro takes the “best ratio” crown if pricing lands as projected, plus the lead on multimodal (video/image reasoning).

What might slip the date

Things that have historically delayed Gemini Pro releases:

  • Safety testing. Deep Think mode requires extended safety testing; previous Pro launches have slipped 1–4 weeks at this gate.
  • Regulatory. EU AI Act compliance audits for frontier models can add weeks.
  • Compute availability. A capacity-driven slip is possible given AI Ultra demand from I/O 2026.
  • Competitive timing. If OpenAI announces GPT-6 or Anthropic ships Claude 5 in early June, Google may delay 3.5 Pro to either pre-empt or out-launch.

TL;DR

  • Target release: June 2026. No specific date announced.
  • Sits at the top of the Gemini 3.5 family above 3.5 Flash (which is already live).
  • Likely $3–5 / $15–30 per 1M tokens — undercutting GPT-5.5 and dramatically cheaper than Claude Opus 4.7.
  • Deep Think mode will be Ultra-exclusive at launch.
  • Best access path: be on Google AI Ultra, watch the Gemini API changelog and AI Studio.

Sources

  • Mashable: “All the Gemini announcements from Google I/O 2026” (May 19, 2026)
  • AP: “Google I/O Gemini developers conference” (May 19, 2026)
  • Engadget: “Everything Google announced at I/O 2026” (May 19, 2026)
  • Google blog: “Gemini 3: Introducing the latest Gemini AI model from Google”
  • Codersera: “Gemini 3.5 Release Date & Status” (May 19, 2026)