Skip to main content
← All news
AI Models

Gemini 3.7 Flash is three weeks after 3.6, at half the 3.6 price

By AgentRiot Editorial

Google’s workhorse Flash cut is GA as `gemini-3.7-flash`. Vendor benchmarks, plus an introductory $0.75 / $3.75 per million tokens through 31 December 2026.

Gemini 3.7 Flash intro pricing
geminigoogleflashpricingagents

Google is shipping Flash the way other labs ship weekly CLIs. Gemini 3.7 Flash arrived 13 August 2026, three weeks after 3.6 Flash. The Gemini API changelog marks it generally available the same day as gemini-3.7-flash.

We already covered the 3.6 split: workhorse, throughput Lite, and a locked-down cyber model. 3.7 is not another split. It is a faster, cheaper refresh of the workhorse.

The price is the cleanest fact

Google says 3.7 Flash’s introductory price is half the original 3.6 Flash cost per million tokens: $0.75 input / $3.75 output, through the end of the year. The API changelog repeats the introductory window as through 31 December 2026. That intro tariff is exactly half the $1.50 / $7.50 list Google published for 3.6 Flash. After 31 December, do not assume the promo holds. The launch post does not name the post-promo price.

The model card is the place for context-window, modality, and any non-intro SKU rows. Do not plan capacity off a launch graphic.

The benchmarks are Google’s

The launch post includes a vendor-reported scorecard against 3.6 Flash on coding and agent evals Google chose. Those figures are Google’s, on Google’s harness, against Google’s previous Flash. Independent scoreboards are not the basis of this piece. Treat the post’s table as the claim to check later, not as a settled ranking.

Google’s prose around that table is the usual workhorse pitch: fewer retries, better first-pass code, tighter UI adherence from a screenshot or design system, more diligent multi-step tool use. That is positioning. The dated intro price is the fact you can invoice against.

Where Google says you can get it

The launch post and API changelog put gemini-3.7-flash on the Gemini API the same day. Google also lists developer surfaces (AI Studio and related IDE/agent tools) and enterprise Gemini apps. On the consumer side it says Gemini Spark, the always-on agent in the Gemini app, starts using 3.7 Flash for Google AI Pro and Ultra subscribers in the countries Google lists. The launch footnote excludes some regions; check that list before you assume Spark already moved.

Safety language in the post is a pointer, not a new eval dump: updated CBRN and cyber-offense safeguards, plus the model card. There is no independent red-team report attached to the launch.

Why this is a story

Three weeks is a cadence, not a generation. The reason to write it is the combination Google is actually selling: a named GA model id, a dated intro price, and Spark already routed onto it. If you priced agents against 3.6 Flash last month, the invoice math changed on 13 August. If you were waiting for a new Flash line, this is not that. It is 3.6’s successor at half the list price, with Google’s own scoreboard attached.

Sources