Gemini 3.8 Flash is half price until January 1
Gemini 3.8 Flash launched at $0.75/$3.75 per million tokens, but Google's own pricing page says the rate doubles on January 1, 2027. Budget for the real number.
Google shipped Gemini 3.8 Flash to general availability on September 2, and the launch price is genuinely cheap: $0.75 per million input tokens, $3.75 per million output. Read one line further down the pricing page and the introductory rate expires. On January 1, 2027 it becomes $1.50 and $7.50 — exactly double, on both sides of the meter (Google pricing).
What actually happened
3.8 Flash is Google's workhorse tier, positioned for long-horizon software engineering, autonomous agents and multi-step enterprise workflows. It carries a 1M-token context window, 64k max output tokens, and three tunable thinking levels — low for latency-critical work like incident response and fast data analysis, medium (the default) for complex code and agentic use, high for deep reasoning and tool orchestration (Google blog).
It is live in the Gemini API through AI Studio and Android Studio, in Gemini Enterprise, and in the consumer surfaces — the Gemini app for AI Pro and Ultra subscribers, AI Mode in Search, and Gemini in Google Sheets. A security-focused Gemini 3.8 Flash Cyber variant was announced the same day.
Why the Gemini 3.8 Flash price cliff matters for your business
Four months is long enough to build a real dependency. You wire an agent to 3.8 Flash in September, it works, the invoice looks fine, and nobody revisits it. Then January arrives and the same traffic costs twice as much — not because you scaled, but because a promotional window closed.
The fix is boring and takes an afternoon. Put the model name in config, not in code. Log token counts per workflow, not just a monthly total, so you can see which job is actually expensive. Then run the math at the January rate, not today's, before you decide a workflow is economical. If a pipeline only pencils out at $0.75, it was never a pipeline — it was a coupon.
Thinking levels are the other lever most teams leave untouched. Medium is the default, and defaults are what ship. Plenty of agent steps — classification, extraction, routing, formatting — get identical results at low effort for a fraction of the output tokens. That is a bigger cost line than the headline rate.
Key takeaways
- Gemini 3.8 Flash launched September 2, 2026 at $0.75/$3.75 per million input/output tokens
- Google's pricing page sets the standard rate at $1.50/$7.50 from January 1, 2027 — a 2x increase on both sides
- 1M-token context, 64k output, and low/medium/high thinking levels; medium is the default
- Available via the Gemini API, AI Studio, Android Studio and Gemini Enterprise; Flash Cyber announced alongside
- Model the January price before you commit a workflow, and tune thinking level per step — that's where the real spend hides
An intro price is not a cost model. We build vendor-agnostic routing so a rate change is a config edit, not a rewrite, and we instrument token spend per workflow so you can see it coming. Run your numbers, or tell us what your agents currently cost.
Sources: Google Developers Blog, Gemini API pricing.
- #gemini
- #llm-pricing
- #ai-agents
- #cost-control
- #model-routing
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Thinking Machines at $40B: bet on what actually shipped
Accel is reportedly in talks to lead a $1B round for Thinking Machines at $40B on $100M revenue. The round is noise. The open weights are the durable part.
Read itMAI-Transcribe-2 drops transcription to $0.10 per hour
Microsoft's MAI-Transcribe-2 launched at $0.10 per hour of audio with 60 languages and diarization. The model is a commodity — your pipeline is the lock-in.
Read it