Gemini 3.7 Flash API
Choose Gemini 3.7 Flash
Google’s newest Flash-tier workhorse for cost-efficient coding, agentic workflows, knowledge work, and multimodal reasoning, with improved token efficiency over Gemini 3.6 Flash and a 1M-token context window.
Gemini 3.7 Flash
Google Flash-tier workhorse model
gemini-3.7-flashProduction coding and full-stack refactoring, multi-step agents and orchestration, document and chart analysis, and high-volume workloads where better token efficiency lowers cost per completed task.
Gemini 3.7 Flash pricing
Estimate a request with the interactive pricing calculator. All user groups use the official Gemini 3.7 Flash rate.
Token calculator
Enter the token mix for one request.Estimated request cost
Gemini 3.7 FlashMinimum charge: 0.01 credits per request.
EvoLink vs Google direct
Same token mix, default group price.Budget guide
Approximate requests using the current token mix.For quick testing
For regular development
For production evaluation
Model pricing
| Model | Context | Input tokens | Cache read tokens | Output tokens |
|---|---|---|---|---|
Gemini 3.7 Flashgemini-3.7-flash | All context sizes | $0.675 / 1M-55% 45.9 cr / 1M$1.500official price | $0.068 / 1M-55% 4.6 cr / 1M$0.150official price | $3.375 / 1M-55% 229.5 cr / 1M$7.500official price |
Gemini 3.7 Flash
All context sizesUSD and credits are shown per 1M tokens. Live backend pricing takes priority over these frozen fallback rates.
Gemini 3.7 Flash API for cost-efficient coding and agent workflows
Call Google’s newest Flash-tier workhorse through EvoLink’s unified API. Gemini 3.7 Flash improves coding, knowledge work, and multimodal reasoning over Gemini 3.6 Flash, plus a 1,048,576-token context window, prompt caching, structured output, and tool use — at Google’s introductory $0.75 / $3.75 per million input / output tokens (through December 31, 2026).
Where Gemini 3.7 Flash earns a place in a production model stack
Google positions Gemini 3.7 Flash as a workhorse: better coding, knowledge work, and multimodal performance with meaningfully improved token efficiency over Gemini 3.6 Flash. Its strongest fit is production work where good-enough reasoning at a lower token cost beats paying for a premium tier.
Cost-efficient coding and refactoring
Gemini 3.7 Flash targets everyday coding, prototyping, and full-stack refactoring with fewer tokens and fewer model calls than Gemini 3.6 Flash. It is available in Google’s Antigravity agent environment. Measure accepted patches and review time, not isolated snippet quality.
Agentic workflows and orchestration
It fits multi-step orchestration, tool selection, structured output, and code execution where lower per-call cost compounds across long agent runs. Preserve complete assistant messages, tool-call IDs, arguments, and tool results across turns.
Knowledge work and multimodal reasoning
Use it for document analysis, chart interpretation, and multi-element web layout generation across a 1M-token context. Retrieval and document structure still matter: a 1M window does not make irrelevant context useful.
When another route is the better choice
Trivial, latency-critical, or very high-volume classification and extraction usually belong on the cheaper Gemini 3.5 Flash-Lite. Route the hardest reasoning to a Pro-tier model. Escalate to 3.7 Flash only when its efficiency actually lowers cost per accepted task.
What early Gemini 3.7 Flash reactions suggest—and what still needs proof
Gemini 3.7 Flash launched on 2026-08-13. Treat launch-day community reactions as hypotheses to verify on your own tasks, tools, budgets, and acceptance criteria—not as settled benchmarks.
Same price as 3.6 Flash, stronger agentic execution
Google positions 3.7 Flash as its most capable workhorse model yet for coding and agents, with major gains in code generation and terminal execution, at exactly the same price as Gemini 3.6 Flash. On the same workload the win comes from fewer retries and shorter agent loops, not from a lower per-token rate.
It is a Flash-tier workhorse, not a top-end engineering model
Some launch-day discussion places it behind larger frontier models on hard software-engineering benchmarks. If your work is deep repository refactoring or long autonomous coding, benchmark it against a Pro-tier or premium model before making it the default.
Verify token usage per task on your prompts
Some launch-day reports observed higher token usage per task than Gemini 3.6 Flash. Check completion length, instruction-following, and tool-call discipline on representative prompts so extra reasoning does not quietly raise output cost.
It is a closed, API-only Google model
Gemini 3.7 Flash has no open weights and cannot be self-hosted. Access is through the Gemini API and platforms such as Google AI Studio, Antigravity, and unified gateways like EvoLink—so route by cost and reliability, not by local deployment.
Why Gemini 3.7 Flash can handle these workloads
Gemini 3.7 Flash is most useful when a large context window, improved token efficiency, and reusable prompt prefixes work together. Context capacity alone does not improve an answer; the workload still needs relevant evidence, clear structure, and an output budget.
A 1M-token workspace, not a target to fill
The 1,048,576-token window can keep related code, specifications, and prior tool results available without excessive chunking. Retrieval and context compaction still matter because irrelevant input competes for attention and increases processing cost.
Improved token efficiency over 3.6 Flash
Gemini 3.7 Flash aims to complete multi-step workflows in fewer turns and fewer tokens. Fewer model calls and fewer total tokens are where the cost advantage comes from, so track total tokens per accepted task rather than a single request price.
Prompt caching pays off when prefixes stay stable
Repository instructions, system prompts, reference material, and tool schemas create the strongest cache opportunity when their ordering stays consistent. Frequent model or prompt-structure changes can force the long prefix to be processed again.
What to verify before routing production traffic to Gemini 3.7 Flash
A suitable workload can still fail because the integration uses the wrong identifier, sends unsupported parameters, or drops agent state between turns. Verify the request surface and conversation contract before evaluating model quality.
Use the exact model ID gemini-3.7-flash
Send model "gemini-3.7-flash" (with dots) on the EvoLink API route. The dashed form gemini-3-7-flash is only the page URL, not the API model parameter.
Use OpenAI Chat Completions or the Gemini native API
EvoLink exposes Gemini 3.7 Flash through OpenAI-compatible /v1/chat/completions and the Gemini native generateContent endpoint. Keep the same EvoLink API key for either protocol.
Mind the breaking parameter changes
Per Google’s migration checklist, remove custom temperature, top-P, top-K, and candidate_count, replace numeric thinking_budget with the thinking_level string, and remove prefilled model turns. Update older Gemini request builders accordingly.
Replay complete assistant and tool state
Multi-turn agents should retain complete assistant messages, tool-call IDs, arguments, and tool results. Keeping only the final text breaks state continuity and can make later steps fail even when the context window is large enough.
Compare cost per accepted task, not token price alone
Gemini 3.7 Flash wins on economics when its efficiency reduces tokens, model calls, retries, or human rework on the same workload. Evaluate identical task sets instead of comparing isolated prompt prices.
If Gemini 3.7 Flash produces usable results with fewer tokens and less review effort, its token efficiency compounds into a real cost advantage. If quality slips on your hardest tasks, escalate those to a Pro-tier model and keep 3.7 Flash for the efficient majority.
Compare leading long-context models after workload testing
EvoLinkFirst verify whether Gemini 3.7 Flash reduces retries and review effort on your tasks. Then compare price, context, caching, and workload fit to choose the production route.
| Model | Gemini 3.7 Flash | Gemini 3.5 Flash Lite | Gemini 3.1 Pro |
|---|---|---|---|
| Input / output | $0.675 / $3.375 | $0.3 / $2.5 | $1.68 / $10.08 |
| Context | 1M | 1M | 1M |
| Caching | Automatic cache reads | Context cache | Context cache |
| Best for | Production coding and full-stack refactoring, multi-step agents and orchestration, document and chart analysis, and high-volume workloads where better token efficiency lowers cost per completed task. | Pro-tier Gemini for the deepest reasoning tasks the Flash line is not built to handle. | Pro-tier Gemini for the deepest reasoning tasks the Flash line is not built to handle. |
Other Gemini models on EvoLink
Gemini 3.5 Flash
The previous-generation Flash workhorse that 3.7 Flash improves on for token efficiency and output cost.
View model
Gemini 3.5 Flash Lite
The cheapest, fastest 3.5-class route for classification, extraction, and high-throughput subagents.
View model
Gemini 3.1 Pro
The Pro-tier step up when a task needs deeper reasoning than the Flash line can deliver.
View model
Gemini 3.1 Flash Lite
The lightweight 3.1 route for high-volume, cost-sensitive, and low-latency workloads.
View modelOther text models
GPT-5.6
OpenAI’s tiered frontier family (Sol/Terra/Luna) for capability, latency, and cost-routing flexibility.
View model
Claude Opus 4.8
Anthropic’s premium baseline for long-running coding agents and judgment-heavy review.
View model
DeepSeek V4
Cost-sensitive open baseline for high-volume coding, reasoning, and agent workloads.
View model
Kimi K3
Moonshot’s long-context reasoning route for repository-scale coding and multi-document work.
View modelRelated reading for production teams
Gemini 3.7 Flash release date & rollout
What launched on August 13, 2026, the gemini-3.7-flash model ID, official pricing, and where the model is available.
Read guide
Gemini 3.7 Flash vs Gemini 3.6 Flash
Same price, three weeks apart — the migration checklist, the behaviour changes to test, and when staying on 3.6 is the right call.
Read guide
Compare the Gemini API family
See how 3.7 Flash, 3.5 Flash-Lite, 3.6 Flash and 3.1 Pro differ on price, context, and workload fit before you route.
Read guide
Gemini 3.5 Flash-Lite API
The cheapest, fastest 3.5-class route for classification, extraction, and high-throughput subagents — escalate hard work to 3.7 Flash.
Read guide
Gemini 3.6 Flash API
The previous-generation Flash model at the same price — compare it before moving production traffic to 3.7 Flash.
Read guideGemini 3.1 Pro API
Step up to the Pro tier when a task needs deeper reasoning than the Flash line can deliver.
Read guideGemini 3.7 Flash API FAQ
Is the Gemini 3.7 Flash API available through EvoLink?
Yes. Gemini 3.7 Flash is available as a production model with live backend pricing and official fallback rates.
What model ID should I use for the Gemini 3.7 Flash API?
Send model "gemini-3.7-flash" (with dots). The dashed gemini-3-7-flash is only the page URL, not the API model parameter.
Which protocols and SDKs work with Gemini 3.7 Flash?
Use the OpenAI-compatible Chat Completions endpoint or the Gemini native generateContent endpoint with the same EvoLink API key. There is no Anthropic Messages route for this model.
Does the Gemini 3.7 Flash API really support the full 1M context?
Yes — the route records 1,048,576 input tokens per Google’s documentation. Smaller limits you may see usually come from a particular Gemini product surface or client configuration, not the API model itself.
How should I use the Gemini 3.7 Flash 1M-token context window?
Keep related code, documents, and tool results together, but use retrieval, stable cached prefixes, and context compaction instead of filling the window by default.
How is Gemini 3.7 Flash different from Gemini 3.6 Flash and 3.5 Flash-Lite?
3.7 Flash is priced the same as 3.6 Flash but posts higher official scores on coding and agentic benchmarks such as FrontierCode and Terminal-bench, per Google. 3.5 Flash-Lite remains the cheaper, faster route for classification and high-throughput tasks.
Is Gemini 3.7 Flash actually cheaper to run if it uses more tokens?
Per-token rates match Gemini 3.6 Flash. Google positions 3.7 Flash to finish tasks in fewer turns, while some launch-day reports observed higher token usage per task. Compare total tokens per accepted task on your own workload rather than the per-request rate.
What breaking changes should I handle when migrating to Gemini 3.7 Flash?
Remove custom temperature, top-P, top-K, and candidate_count from generation configs, replace numeric thinking_budget with the thinking_level string, and remove prefilled model turns — all per Google’s official migration checklist.
What happened to the minimal thinking level in Gemini 3.7 Flash?
Google removed it: available levels are low, medium (default), and high, and sending minimal to the Gemini API returns an error. EvoLink is set to downgrade reasoning_effort none or minimal to low so migrated requests do not fail. For high-volume classification pipelines, Gemini 3.5 Flash-Lite is the lower-cost floor.
How are Gemini 3.7 Flash thinking tokens billed?
Thinking tokens are billed at the output rate under Google’s pricing, so verbose reasoning can make bills noticeably larger than the visible answer. Set task-appropriate thinking levels and monitor reasoning plus final-answer tokens.
Can Gemini 3.7 Flash analyze video and audio?
Yes. The model accepts text, image, video, audio, and PDF input and outputs text, which makes it a common choice for video and audio understanding workloads. Image, audio, and live-stream generation are not supported.
Is there a Gemini 3.7 Pro?
No. As of the 3.7 Flash release, Google’s newest Pro-line model remains gemini-3.1-pro-preview, and no 3.5 or 3.7 Pro has been announced on official channels.
How long does the Gemini 3.7 Flash introductory price last?
Google lists $0.75 input and $3.75 output per million tokens as an introductory price through December 31, 2026, with standard rates of $1.50 and $7.50 from January 1, 2027. See the pricing section on this page for the current EvoLink rate.
Is Gemini 3.7 Flash more reliable than 3.6 Flash at following instructions?
Google reports similar safety and tone to 3.6 Flash, while one independent tracker measured a higher hallucination rate at launch. Verify instruction-following and factual discipline on your own representative tasks before promoting it to production.
What should a production Gemini 3.7 Flash evaluation measure?
Track first-pass success, accepted deliverables, retries, output tokens, cache hits, valid tool calls, time to accepted result, human correction, and fallback rate.