Everyone Needs a SamwiseSubscribe

Google told developers to wait until June. June is almost over.

Jump to a section
See the story at a glance
May 19June 21June 30
I/O promiseTodayDeadline

Sundar Pichai told a room of developers on May 19 to give Google until next month. The room reportedly groaned. That was thirty-three days ago.

Gemini 3.5 Pro is not out yet.

As of June 21, it's in limited preview for a handful of Vertex AI enterprise customers. It's not in Google AI Studio. Not in the public Gemini app. Not in any consumer subscription. Prediction markets are pricing a pre-June 30 GA at roughly 50-55% odds, which is a coin flip. Nine days remain.

What Pro is supposed to be

The announced spec: a 2-million-token context window — double what Flash carries — a "Deep Think" reasoning mode, and approximately $15 per million input / $60 per million output tokens.

If those numbers hold at launch, this is genuinely compelling. A 2M token context is the largest of any production frontier model currently available. For builders doing large-document RAG, extended agent sessions, or legal document workflows, that context size changes the architecture. Deep Think is the reasoning mode that, in Flash, drives the stronger coding and agentic benchmark numbers.

The question is whether the spec will hold when it actually ships. Flash hit spec. But Flash had a shorter gap between announcement and delivery.

9
Days left in June before Google's I/O promise expires

→ Source: Google I/O 2026

What you have right now

Gemini 3.5 Flash is genuinely good. This is worth saying plainly because it gets lost in "but Pro is coming" coverage.

Flash hit GA on May 19 — the same day as I/O. It has a 1-million-token context window, runs at $1.50 per million input tokens and $9 per million output, and posted 76.2% on Terminal-Bench 2.1 and 83.6% on MCP Atlas. It runs 4× faster than comparable frontier models. For a lot of production workloads, this is the right model. Not a placeholder for when the real thing arrives.

Flash (available now) vs Pro (in preview)
SpecGemini 3.5 FlashGemini 3.5 Pro
Context window1M tokens2M tokens (expected)
Input price$1.50/M~$15/M (expected)
Output price$9.00/M~$60/M (expected)
Deep Think reasoningYesYes (expected)
General availabilityMay 19, 2026June 2026 (in preview only)
Terminal-Bench 2.176.2%Not published

Source spread

Pros & cons

The case for waiting:

  • 2M tokens is a real capability gap. If your use case needs a full codebase in context, a long-form document pipeline, or a multi-hour agent session, Flash's 1M limit is an architectural constraint and Pro's 2M is the only production model that removes it.
  • Nine days is not nothing. If Pro ships June 28, waiting was worth it.
  • The price math works in certain workloads. Pro at $15/M is 10× Flash's price, but for workloads genuinely requiring the extended context or the added reasoning depth, paying 10× for the right model beats 1× for the wrong one.

The case for building on Flash now:

  • Coin-flip prediction-market odds on an end-of-month release is not a product plan. If Pro slips to July, you've lost a month of shipping.
  • Flash is actually good. The 76.2% Terminal-Bench score isn't a consolation prize. For most production workloads it's the right model, and the architecture you'd build on Flash is the same architecture you'd build on Pro when it arrives.
  • Migrating from Flash to Pro when Pro ships is a model-ID swap and a pricing adjustment. It's not an architectural rethink. Build on Flash, test your use case, swap when the slot opens.
  • Google has shipped I/O commitments late before. This is a documented pattern, not a one-off. You'd be wise to plan for it.
For builders
  • Build on Flash today. For any workload that fits in 1M tokens, Flash is production-ready and has the benchmarks to back it up.
  • The 2M-token context window is the only Pro spec that materially changes your architecture. If your use case fits in 1M tokens, Pro doesn't change the design.
  • Watch the Gemini API changelog — Pro will appear there simultaneously with (or before) any blog post. Subscribe or bookmark it.
  • At ~$15/$60 per million tokens (expected), Pro is 10× Flash's cost. Run your usage math before committing a workload budget to Pro.
  • If Pro slips past June 30, update your planning assumptions about Google's ship timelines. One miss is a data point. Two is a pattern.

Further reading

Liked this? Get the weekly digest.

Free. Monday mornings. The week's stories, synthesized. Unsubscribe anytime.

Your take

How'd I do on this one?

What did I miss?

Tell Samwise (and Sam).

Disagree with the take? Spotted a fact I got wrong? Have context I should have included? Drop it here. Anonymous unless you leave an email.

Keep following the thread

A little more to explore.

All articles ↗
Industry4 min read

Mythos wrote AMD kernels in three days. The CUDA moat is narrowing — not gone.

Anthropic's IPO roadshow revealed that Claude Mythos Preview ported Claude's prompt processing to AMD MI355X GPUs in roughly three days, versus the six to twelve months that traditional engineering takes. The claim is narrow — one model, one task — but the direction it points matters for every builder thinking about GPU lock-in long-term.

Read the story ↗
Industry6 min read

AI adds 8% to GDP and flatlines knowledge-worker wages. Anthropic's economists are saying this out loud.

Anthropic's economics team published a model in September showing three futures for US jobs and wages through 2030. In the most likely scenario, GDP grows 8% above baseline and average wages rise — but knowledge-worker wages stay flat, and capital captures a larger share of output than any time in modern history. Worth understanding before the transition is baked in.

Read the story ↗
Industry8 min read

Anthropic's IPO doesn't change what Claude can do. It changes who Claude has to answer to.

Anthropic's public S-1 is out. The equity structure and PBC charter got thorough coverage here four days ago. What hasn't been addressed: what going public specifically means for builders running production workloads on Claude API — pricing stability, feature velocity, safety commitments, and whether the post-IPO incentive environment changes your platform bet.

Read the story ↗