"Gemini 3.7 Flash" Pricing Leak: Unverified X Rumor, Explained

A chain of X posts claims Google is shipping "Gemini 3.7 Flash" today, August 13, 2026, at half the price of Gemini 3.6 Flash. Google has not confirmed this, and no reputable outlet has corroborated it. This post is speculative coverage of a leak, not a report on an official release — treat everything below the confirmed-pricing section as unverified until Google says otherwise, and expect this page to get updated or retracted once real information lands.

The claim comes from accounts including @synthwavedd, @LuminaBench, @AiBattle_, and @cheatyyyy, all repeating the same numbers: $0.75 per million input tokens, $3.75 per million output tokens, and a same-day launch. The same thread claims Google has internally cancelled Gemini 3.5 Pro in favor of a larger "Gemini 4," and that Sergey Brin is pushing resources into recursive self-improvement (RSI). None of it has a primary source attached — no model card, no blog.google post, no Gemini API changelog entry.

explainx.ai has covered Gemini leak chains before, including the July 13 Gemini 3.5 Pro benchmark leak that promised a July 17 launch and then quietly slipped into "testing with partners" status by the time Gemini 3.6 Flash actually shipped on July 21. The pattern is worth knowing before reacting to this one.

TL;DR — what people are asking

What the leak actually claims

Stripped to bullet points, the X chain circulating on August 13, 2026 says:

• Google will release "Gemini 3.7 Flash" later today
• Pricing: $0.75 per million input tokens / $3.75 per million output tokens
• That's roughly half of Gemini 3.6 Flash's confirmed $1.50/$7.50 pricing
• Gemini 3.5 Pro has been internally cancelled, with effort redirected to a larger "Gemini 4" model
• Sergey Brin is personally pushing resources into recursive self-improvement (RSI)

No screenshot of an internal document, no model card, no API string, and no link to blog.google accompanies any of these posts. That's the same shape as the July 13 Gemini 3.5 Pro leak from @EntelligenceAI, which also carried a specific number (a July 17 launch date) with no primary source — and which slipped by four days before Google shipped something different than what was rumored.

The real, confirmed baseline: Gemini 3.6 Flash

Here's what is actually true, sourced from Google's own July 21, 2026 announcement and independently reported by 9to5Google:

Gemini 3.6 Flash's real move was a 17% output-price cut against 3.5 Flash — not a halving. If the $0.75/$3.75 numbers in today's leak turn out accurate, that would be a much steeper single-generation price cut than Google's last two Flash releases delivered, which is itself a reason for skepticism: explainx.ai's coverage of the 3.6 Flash launch found Hacker News testers reporting mixed-to-negative results on Google's own token-efficiency claims for that release, let alone a rumored doubling of price efficiency on top of it.

Why the "3.5 Pro cancelled" claim deserves the same skepticism

This part of the leak overlaps with a thread explainx.ai has already tracked twice. The timeline as confirmed by Google itself:

1. July 13, 2026 — a leak claimed Gemini 3.5 Pro would launch July 17, beating Claude Fable 5 and GPT-5.6 in internal evals
2. July 21, 2026 — Google shipped 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber instead, with 3.5 Pro described only as "testing with partners" — no cancellation language, no GA date
3. Same announcement — Google confirmed it had started pre-training on a larger Gemini 4, reported separately in explainx.ai's coverage of Google's Frozen v2 chip

So two pieces of today's rumor rest on real prior facts — 3.5 Pro is delayed, and Gemini 4 pre-training is real — stitched together with a new, unsourced claim ("cancelled") that goes further than anything Google has said on the record. That's the exact pattern GEO and news-literacy practice flags: a rumor gains credibility by wrapping a confirmed fact around an unconfirmed one. Don't let the true parts vouch for the false part.

The Sergey Brin / RSI claim, separated from fact

Recursive self-improvement — models that meaningfully speed up their own training, evaluation, or research loop — is a real and heavily discussed industry theme in August 2026. explainx.ai has covered it from multiple angles this summer, including Weco AIDE²'s Level 1 RSI ladder result and the broader AGI-to-ASI pathways research from DeepMind. That context makes an RSI-flavored rumor about Google plausible-sounding on its face.

But plausibility isn't confirmation. The specific claim — that Sergey Brin is personally redirecting resources toward RSI as part of this leak chain — has no attached statement, interview, or internal memo. It's riding on the credibility of a real trend, the same way the "3.5 Pro cancelled" claim rides on a real delay. Treat it as a separate, unverified assertion until Google or Brin himself says something on the record.

What a real Flash price halving would mean for builders — if it happens

Even framed as unconfirmed, the rumor is worth thinking through, because Google's Flash-tier cadence this year has been genuinely fast: Gemini 3.5 Flash (June) → Gemini 3.5 Flash-Lite (July 21) → Gemini 3.6 Flash (July 21) → a rumored "3.7 Flash" today, six weeks later. If Google is iterating Flash-tier pricing roughly monthly, that has real implications regardless of whether today's specific numbers hold up:

• High-volume agentic workflows are the target. Flash-tier pricing wars are won on throughput jobs — document extraction, agentic search, classification — not frontier reasoning. A further halving of output price would matter most to teams already running Flash-class models at scale, the same audience explainx.ai flagged benefiting from 3.5 Flash-Lite's 350 tokens/second throughput.
• Per-token price cuts don't automatically mean lower total cost. Independent Hacker News testing of Gemini 3.6 Flash found some workloads got less token-efficient despite the lower sticker price. Any team evaluating a real "3.7 Flash" (or any future Flash release) should re-run a private cost-per-task harness rather than trusting a headline per-token number.
• Rapid Flash iteration signals where Google's actual competitive pressure is. Three Flash-tier releases in six weeks against a stalled Pro tier suggests Google is defending the high-volume, cost-sensitive segment of the market against GLM-5.2 and DeepSeek V4 pricing, not chasing frontier reasoning parity with Claude Fable 5 or GPT-5.6 right now.

What to do until Google confirms or denies this

1. Don't change production routing today. Confirmed pricing is still Gemini 3.6 Flash at $1.50/$7.50 and 3.5 Flash-Lite at $0.30/$2.50 — nothing else is real yet.
2. Watch blog.google and the Gemini API changelog, not X reply counts, for the actual announcement.
3. If "Gemini 3.7 Flash" does ship, run your own harness before trusting either the leak's numbers or Google's own marketing chart — see the same lesson from 3.6 Flash's mixed independent benchmarks.
4. Treat the "3.5 Pro cancelled" and Brin/RSI claims as separate, weaker rumors riding on the same thread as the pricing leak — neither has independent sourcing.

This post will be updated with a dated banner once Google confirms, denies, or the rumor is superseded by an actual announcement.

Related reading

• Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber: what actually changed
• Gemini 3.5 Pro benchmark leak — July 13, 2026
• Google's Frozen v2 chip and the start of Gemini 4 pre-training
• Gemini hit 1 billion users — but it's not the same billion as ChatGPT
• Weco AIDE² — Level 1 recursive self-improvement, 8 days, 7 agent versions
• DeepMind's AGI-to-ASI paper: four pathways
• Google Gemini 3.5 complete guide
• Official: Google Gemini API changelog · 9to5Google's Gemini 3.6 Flash launch report

Sources: X posts from @synthwavedd, @LuminaBench, @AiBattle_, @cheatyyyy as of August 13, 2026 — unverified, no primary Google source attached.

This post covers an unconfirmed leak, not an official Google announcement. Pricing, cancellation, and RSI claims attributed to the X leak chain are not verified by Google, Bloomberg, TechCrunch, or any other reputable outlet as of publication. Confirmed pricing (Gemini 3.6 Flash, 3.5 Flash-Lite) reflects Google's July 21, 2026 announcement. Check blog.google and the Gemini API changelog before making any production decisions, and expect this page to be updated or retracted once official information is available.