A new coding or agent workflow
Pilot Gemini 3.7 Flash first.
Google now gives 3.7 the workhorse role. Use repository tasks, tool calls, and acceptance tests rather than relying on provider benchmarks alone.
Google model field guide · checked August 14, 2026
Gemini 3.6 Flash still matters—but 3.7 is the newer workhorse.
Google still lists Gemini 3.6 Flash with no retirement date announced, but now positions 3.7 Flash as its newest workhorse for coding and agents. Keep 3.6 as a valid existing-workflow baseline; for new work, test 3.7 first and compare accepted-task cost.
Provider record
Provider role
July 2026 Flash model for coding, knowledge work, multimodal tasks, and agentic workflows. Gemini 3.7 Flash is Google's newer workhorse.
Current listed price
$0.75 input and $3.75 output per million tokens through December 31, 2026; $1.50 and $7.50 starting January 1, 2027.
Lifecycle status
Google lists gemini-3.6-flash as a short-term availability model with no retirement date announced as of August 14, 2026.
Flowith access
Not asserted here. The live Flowith workspace selector is the access authority.
Choose by workload, then run a pilot
An existing 3.6 production workflow
Keep 3.6 as the migration baseline.
Run matched prompts and tools on 3.6 and 3.7, then compare accepted outputs, retries, latency, tokens, and reviewer effort before switching.
High-volume classification or extraction
Include Gemini 3.5 Flash-Lite.
Compare accepted-output cost, tail latency, retries, and review effort.
Record output tokens, tool calls, retries, wall time, accepted results, and human review. Google's benchmark gains are provider evidence, not a guarantee for your workload; a cheaper token is not cheaper if the workflow needs more loops or correction.
Short-term availability is a planning signal
Google says short-term models retire 45 days after a replacement is released. Its current table does not name a 3.6 replacement or retirement date. Recheck that table before release planning instead of treating today's blank date as a durability promise.
Related decisions
Official sources
- Google 3.6 launch announcement — original model role, launch price, access, and provider evaluations.
- Google 3.7 launch announcement — newer workhorse role, comparison claims, and dated introductory pricing.
- Google Cloud pricing — current 3.6 and 3.7 price windows and consumption terms.
- Google model lifecycle table — current short-term availability and retirement status.
- Gemini 3.6 Flash model card — current model and safety details.
Gemini 3.6 Flash FAQ
What is Gemini 3.6 Flash?
Gemini 3.6 Flash is Google's July 2026 Flash model for coding, knowledge work, multimodal analysis, and agentic workflows. Google introduced 3.7 Flash in August as its newer workhorse.
How much does Gemini 3.6 Flash cost?
Google currently lists $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026, followed by $1.50 input and $7.50 output starting January 1, 2027. Check the live pricing table and applicable inference mode before budgeting.
Is Gemini 3.6 Flash being retired?
Google's model lifecycle table lists Gemini 3.6 Flash in its short-term availability group with no retirement date announced as of August 14, 2026. That current status is not a guarantee of indefinite availability, so keep the model ID configurable and monitor the live table.
Should I use Gemini 3.6 Flash or 3.7 Flash?
Pilot 3.7 first for a new coding or agent workflow because Google positions it as the newer workhorse. For an existing 3.6 workflow, compare both versions on the same acceptance set before migrating; a newer release does not prove a better result for every workload.
Is Gemini 3.6 Flash available in Flowith?
This page does not claim Flowith availability. Verify the current Flowith workspace model selector before designing a workflow.