High-throughput coding agents
Agent loops that need long context, function calls, code execution and lower latency than a large flagship model.
Independent tool overview
Gemini 3.5 Flash at a glance
Gemini 3.5 Flash is Google's stable multimodal, text-output model for fast coding and agentic workloads, with a one-million-token input window, tool use and paid API rates of $1.50 input and $9 output per million tokens.
Visit the official Gemini 3.5 Flash site ↗
- Model ID
- gemini-3.5-flash
- Current status
- Stable and priced; Gemini 3.7 Flash is the recommended migration target
- Inputs
- Text, image, video, audio and PDF
- Output
- Text only
- Context window
- 1,048,576 input tokens
- Maximum output
- 65,536 tokens
- Knowledge cutoff
- January 2025
Overview
What Gemini 3.5 Flash is
Gemini 3.5 Flash remains listed as a stable Gemini API model and is available through Google AI Studio and Google's enterprise and developer surfaces. It accepts text, images, video, audio and PDFs, then returns text. The model is designed for coding, multimodal understanding and multi-step agent loops rather than image, audio or real-time voice generation.
There is now a newer migration target. Google's current documentation recommends Gemini 3.7 Flash for workloads moving from 3.5 Flash and says the upgrade improves coding, spatial and multimodal reasoning, design adherence and agent reliability. Existing 3.5 integrations do not need to be labeled inactive, but new projects should benchmark 3.7 before locking 3.5 into production.
Use cases
Who Gemini 3.5 Flash is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Multimodal document analysis
Applications reasoning across text, screenshots, video, audio and PDFs while producing structured text.
Grounded search workflows
Paid applications that need Google Search, Maps, URL context or file search integrated with a reasoning model.
Existing 3.5 production systems
Teams maintaining a stable model ID while they test output quality, latency and compatibility against Gemini 3.7 Flash.
Capabilities
Core Gemini 3.5 Flash features
One-million-token context
Accepts up to 1,048,576 input tokens and can return as many as 65,536 output tokens.
Multimodal understanding
Processes text, images, video, audio and PDF documents in one model while returning text.
Thinking
Can spend internal reasoning tokens on harder tasks; those tokens are included in the paid output-token charge.
Function calling and structured output
Supports tool schemas and machine-readable responses for application and agent orchestration.
Code execution and computer use
Supports code execution and offers computer use in preview for workflows that need to act rather than only answer.
Search and Maps grounding
Paid API workflows can connect responses to Google Search and Maps, with separate usage charges after the included allowance.
File search and URL context
Can retrieve from indexed files and supplied URLs as part of a longer reasoning or agent workflow.
Batch, Flex and Priority inference
Offers discounted asynchronous or flexible capacity and a higher-priced priority service in addition to standard inference.
Process
How the Gemini 3.5 Flash workflow works
Step 1
Decide whether to start on 3.7
For a new system, benchmark Gemini 3.7 Flash first because Google now names it as the upgrade target from 3.5.
Step 2
Build a representative evaluation set
Include real prompts, difficult edge cases, tool failures, long-context examples and expected structured outputs.
Step 3
Set the reasoning and tool budget
Cap output, agent steps, searches and retries because thinking tokens and repeated tool calls affect both latency and cost.
Step 4
Ground only when needed
Use Search, Maps, file retrieval or URL context for claims that require current or private evidence, then preserve source attribution.
Step 5
Validate every action
Check schemas, permissions, code execution and computer-use outcomes before allowing the model to affect shared or production systems.
Step 6
Run a controlled 3.7 migration
Remove deprecated sampling parameters and prefilled model turns, compare behavior and cost, then roll forward behind monitoring and rollback.
Cost
Gemini 3.5 Flash pricing and free plan
Gemini 3.5 Flash has a free developer tier and four paid consumption modes. Standard costs $1.50 per million input tokens and $9 per million output tokens including thinking; Batch and Flex halve inference rates, while Priority charges 1.8 times Standard.
Free tier
$0 within limits
Limited developer access and free Google AI Studio testing.
- Input, output and eligible context caching are free within rate limits
- Search and Maps grounding are not available through the free API tier, but can be tested in AI Studio
- Free-tier content may be used to improve Google products
Standard
$1.50 input / $9 output
On-demand paid inference per one million tokens.
- Output price includes thinking tokens
- Context cache reads cost $0.15 per million tokens
- Cache storage costs $1 per million tokens per hour
- Paid-tier content is not used to improve Google products
Batch
$0.75 input / $4.50 output
Discounted asynchronous processing per one million tokens.
- No free Batch tier
- Context cache reads cost $0.075 per million tokens
- Best for delay-tolerant bulk work
Flex
$0.75 input / $4.50 output
Discounted flexible-capacity inference per one million tokens.
- No free Flex tier
- Context cache reads cost $0.08 per million tokens
- Availability and latency tradeoffs should be load-tested
Priority
$2.70 input / $16.20 output
Higher-priced capacity for workloads that need priority service.
- Output price includes thinking tokens
- Context cache reads cost $0.27 per million tokens
- Cache storage remains $1 per million tokens per hour
Search and Maps grounding
5,000 free, then $14/1,000 queries
Separate paid-tool allowance shared across Gemini 3.x models.
- Allowance is monthly and shared
- Billing is per search query, not necessarily per user prompt
- One model request can trigger multiple billable search queries
Pricing checked . Check current pricing at the source ↗
Assessment
Gemini 3.5 Flash strengths and limitations
Where it stands out
- Combines a large context window with broad multimodal input and relatively fast Flash-series inference.
- Supports core agent primitives: thinking, function calls, structured output, code execution and retrieval.
- Offers several consumption modes for trading latency and availability against cost.
- Paid-tier prompts are not used to improve Google products under the published pricing terms.
- The stable model ID remains usable while teams evaluate a controlled migration.
What to consider
- Gemini 3.7 Flash is now Google's recommended migration target from 3.5 Flash, so 3.5 is no longer the obvious choice for a new integration.
- The model returns text only; it does not generate images or audio and does not support the Live API.
- Its January 2025 knowledge cutoff requires grounding for later facts.
- Computer use is a preview capability and needs stronger permissions, isolation and human approval than ordinary text generation.
- Thinking tokens are billed as output, so agentic reasoning can cost more than the visible answer suggests.
- A million-token window does not guarantee reliable recall or reasoning over every detail in a very large prompt.
- Search grounding charges each generated query after the shared allowance, and one request can issue more than one search.
- Free-tier content is used to improve Google products according to the API pricing table; sensitive production data belongs on an appropriate paid service.
- Migrating to Gemini 3.7 requires removing deprecated temperature, top-p, top-k and prefilled model-turn patterns.
- Google's launch performance and benchmark claims are vendor results and should be validated on the buyer's own task set.
Compare
Gemini 3.5 Flash alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Consumer
Claude Sonnet 4.6
Choose Claude Sonnet 4.6 to compare long-context coding, agent behavior and a different safety and platform stack.
Explore Claude Sonnet 4.6 →Consumer
Grok 4.6
Choose Grok 4.6 for another agentic model with a large context window and integrated SpaceXAI tool ecosystem.
Explore Grok 4.6 →Business Operations
ChatGPT
Choose ChatGPT when the priority is a broad end-user workspace rather than direct model integration.
Explore ChatGPT →Questions
Gemini 3.5 Flash FAQs
Is Gemini 3.5 Flash still available?
Yes. Google currently lists gemini-3.5-flash as a stable model with active free and paid pricing.
Has Gemini 3.5 Flash been replaced?
It has not been removed, but Gemini 3.7 Flash is Google's recommended migration target for Gemini 3.5 Flash workloads and is the better starting benchmark for new projects.
How much does Gemini 3.5 Flash cost?
Standard paid pricing is $1.50 per million input tokens and $9 per million output tokens, including thinking tokens. Batch and Flex are $0.75 input and $4.50 output; Priority is $2.70 input and $16.20 output.
Is there a free tier?
Yes, within published rate limits. Google says free-tier content may be used to improve its products, while paid-tier content is not.
What is the Gemini 3.5 Flash context window?
The model supports up to 1,048,576 input tokens and 65,536 output tokens.
What inputs and outputs does it support?
It accepts text, images, video, audio and PDFs and returns text. It does not generate images or audio and does not support the Live API.
Does Gemini 3.5 Flash support tool use?
Yes. It supports function calling, structured outputs, code execution, file search, URL context, Google Search and Maps grounding, plus computer use in preview.
How much does Google Search grounding cost?
Paid Gemini 3.x models share 5,000 free search requests per month, then charge $14 per 1,000 queries. A single model request can produce multiple billable search queries.
What changes are needed for Gemini 3.7 Flash?
Google says migrations from 3.5 Flash should remove deprecated temperature, top-p and top-k sampling parameters and prefilled model turns, then validate the changed behavior.
Should a new application use 3.5 or 3.7 Flash?
Start by evaluating Gemini 3.7 Flash because Google recommends it as the migration target. Keep 3.5 only when its current behavior, availability or an incomplete migration gives it a measured advantage.
Bottom line
Our Gemini 3.5 Flash verdict
Gemini 3.5 Flash is still a capable, stable model for multimodal and agentic systems, but it now sits in a transition window. Existing teams can keep it while they run a proper regression and cost test; new teams should begin with Gemini 3.7 Flash. The biggest implementation risks are hidden thinking and search costs, preview computer use, free-tier data handling and assuming a huge context window removes the need for retrieval and evaluation.
Visit Gemini 3.5 Flash website ↗Similar Consumer Tools
View all Consumer AI tools →
Alexa for Shopping
Alexa for Shopping - Amazon's new shopping agent for Q&A, price tracking, and Auto-Buy across devices
Qwen3.7-Max
Qwen 3.7 Max- Alibaba's new flagship model for long-horizon agentic tasks
Incognito Chat
Incognito Chat - Meta’s new way to have private conversations with Meta AI via WhatsApp
Claude Opus 4.8
Claude Opus 4.8 - Anthropic's new top model with improvements to reliability, coding, and agentic flows.
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.