Gemini 3.8 Flash — turn context into useful answers
Gemini 3.8 Flash accepts text, images, audio, video and PDFs. Build a conversation around the evidence you have, choose how much thinking to request, and return plain text or structured data. $0.60 / 1M input · $3.00 / 1M output.
- 1M-token input
- 65,536-token output
- Three thinking levels
- Function calling
- Multimodal input
Playground
Explore Gemini 3.8 Flash with native requests. Select an available group and inspect your settings. Live capability verification is pending.
gemini-3.8-flash$0.60 in · $3.00 out per 1M tokensMore Gemini parameters
Try one of these prompts
Messages are billed from your balance at the rates below, like an API call. Failed requests are free.
Gemini 3.8 Flash pricing
Gemini 3.8 Flash rates below reflect current billing settings for each available account group. Compare input, output and cached-input prices before submitting. Standard token rates do not change at the 200,000-token boundary.
Example sizes are for illustration. Your bill follows the tokens each request reports. Standard processing rates, with no 200,000-token price step. Cached tokens are not billed again as uncached input. Tool charges and cache storage are separate from these token examples.
| Category | googlellmPrice · 1M tokens | googlellm-vertextPrice · 1M tokens |
|---|---|---|
| Input | $0.60 | $0.75 |
| Output, including thinking | $3.00 | $3.75 |
| Cached input | $0.06 | $0.075 |
What is Gemini 3.8 Flash?
Gemini 3.8 Flash is a stable Google model for reasoning across text and media. Supply a specific question, supporting material and expected answer format.
Use gemini-3.8-flash. Gemini 3.8 Flash defaults to medium thinking and supports low and high. Its input and output limits match Gemini 3.1 Pro; generation settings and pricing differ.
Official model limits checked October 17, 2026. Input and output have separate limits.
Start with Gemini 3.8 Flash
Start with a task you can judge. Save the input, settings and response.
Open the Playground
Try a Gemini 3.8 Flash message with supporting material. Inspect the request JSON and official controls for thinking levels, output limits and schema settings. Custom JSON is sent unchanged for the Gemini API to validate.
Select an available group
Choose among the Gemini 3.8 Flash groups available to your account. The form updates token rates for the selected group; new groups appear through the same account-aware selection.
Connect your application
Read the Gemini 3.8 Flash API reference for native examples, response fields and complete schemas. Keep the exact model ID in the endpoint when moving a request into your application.
What Gemini 3.8 Flash can help you build
Define the output, then provide enough evidence to check the answer.
Extract fields from documents
Ask Gemini 3.8 Flash for a JSON object with fields you define. Include source references where possible, and verify the extracted values against the document rather than trusting valid syntax alone.
Explain code with context
Give Gemini 3.8 Flash a function, its callers and the observed failure. Request an explanation tied to those lines, then check any proposed correction with your existing tests.
Use functions deliberately
Gemini 3.8 Flash can request a function call using your declared argument schema. Your application remains responsible for validating arguments, executing permitted actions and returning the result.
Read media alongside a question
Put an image, supported recording or PDF in the same Gemini 3.8 Flash conversation as your question. Ask for observable details and have the answer distinguish uncertainty from evidence.
Practical Gemini 3.8 Flash tasks
Suggested tasks to evaluate; no benchmark or customer outcome is claimed.
Classify incoming documents
Give Gemini 3.8 Flash a document and a finite set of categories. Ask for the category, supporting passage and a reason to request human review when the evidence is incomplete.
Review a support conversation
Ask Gemini 3.8 Flash to extract the reported issue, attempted fixes and unresolved question. Preserve the source conversation so a teammate can verify the summary before acting.
Compare a diagram with instructions
Supply a diagram and the instructions it illustrates. Gemini 3.8 Flash can describe apparent inconsistencies; use a focused question so each observation can be checked against the material.
How to use Gemini 3.8 Flash
- 01
Prepare your account
Sign in, add credit and create an API key for server-side use. Open Gemini 3.8 Flash to review available groups and current pricing.
- 02
Assemble the conversation
Use contents for user messages, previous model parts and tool results. Give Gemini 3.8 Flash the relevant context and correct MIME types for media.
- 03
Set the output contract
Choose low, medium or high thinking for Gemini 3.8 Flash. Set an output cap and add a JSON schema when the application requires structured values.
- 04
Inspect the result
Check Gemini 3.8 Flash candidates, finishReason and usageMetadata. Preserve complete content parts and thought signatures when continuing the same conversation.
The native Gemini 3.8 Flash API
Gemini 3.8 Flash uses native generateContent and streamGenerateContent endpoints. Keep request configuration, conversation history and response handling consistent between the Playground and your own application.
Use the stable identifier
The URL contains gemini-3.8-flash. Gemini 3.8 Flash does not require a preview suffix, a thinking-level suffix or an alternate name for streaming responses.
Keep supported settings
Gemini 3.8 Flash uses thinking and output controls rather than configurable temperature, topP, topK or candidateCount. Those fields are not offered as official controls; caller-supplied JSON still passes through unchanged.
Read cumulative usage
Gemini 3.8 Flash reports input, cached input, output and thinking usage. For streaming, use the latest counters; adding cumulative values from every event would count tokens repeatedly.
Gemini 3.8 Flash vs Gemini 3.1 Pro
Compare the documented contracts before choosing a test configuration. Gemini 3.8 Flash and Gemini 3.1 Pro need representative task results to establish which works better for your application.
| Setting | Gemini 3.8 Flash | Gemini 3.1 Pro |
|---|---|---|
| Release status | Stable | Preview |
| Default thinking | medium | high |
| Input limit | 1,048,576 tokens | 1,048,576 tokens |
| Output limit | 65,536 tokens | 65,536 tokens |
| Sampling controls | Temperature, topP, topK and candidateCount ignored | Model-specific sampling fields accepted |
Compare the same task with explicit thinking and output limits. The guide below explains the differences; no measured quality or latency winner is claimed. Google model documentation
Gemini 3.8 Flash limits
Separate input from output
Gemini 3.8 Flash accepts 1,048,576 input tokens and allows 65,536 output tokens. Thinking consumes output tokens, so visible answer length alone does not establish usage.
Generate text, analyze media
Gemini 3.8 Flash produces text. Images, audio, video and PDFs are input material; accepting those formats does not mean this model generates matching media outputs.
Keep the final turn valid
End a Gemini 3.8 Flash request with user content, including a function result when continuing tool use. Do not append a model-prefill turn. Preserve signatures and original tool parts in history.
Verify advanced resources
Gemini 3.8 Flash documents caching and built-in tools. Resource access, media and tool behavior still require live verification for the selected account group.
Use Gemini 3.8 Flash with SeedRouter
One balance across models
Use your SeedRouter balance for Gemini 3.8 Flash and other listed models. Each keeps its own request contract, model identifier and supported output types.
Current rates in context
Review Gemini 3.8 Flash prices beside the model settings. The Playground selects from your available groups and uses the corresponding rate for token-cost calculations.
Clear failed-request billing
Failed Gemini 3.8 Flash requests are not charged. For successful requests, inspect reported usage; output includes thinking, and cached input has its own rate.
Explore other text models
Compare available text models by their own documented capabilities and current rates.
Call GPT-6 Astra with the OpenAI SDK. OpenAI's most capable model, with a 1.05M-token context, live per-token pricing and a playground you can try right now.
View pricingCall GPT-6 Sol with the OpenAI SDK. 1.05M-token context, reasoning effort from none to max, live per-token pricing and a playground you can try right now.
View pricingCall GPT-6.1 Sol with the OpenAI SDK: near-Astra coding and agent work at a fifth of the price, 1.05M-token context, 95% off cached input. Try it now.
View pricingCall GPT-6 Luna with the OpenAI SDK. The lowest-cost GPT-6 model, with a 1.05M-token context, live per-token pricing and a playground you can try right now.
View pricingTry Claude Haiku 5.5 for classification, extraction and summaries. See live token prices, test your prompts, and connect your app with the Messages API.
View pricingTry Claude Sonnet 5.5 for code reviews and document questions. See live token prices, test a prompt, and check verified capabilities before connecting your app.
View pricingGemini 3.8 Flash guides
Compare Gemini 3.8 Flash with Gemini 3.1 Pro using documented settings and a task-specific evaluation.
Gemini 3.8 Flash FAQ
Which Gemini 3.8 Flash model ID should I use?+
Gemini 3.8 Flash uses the stable ID gemini-3.8-flash. Thinking level belongs in generationConfig, not an invented model suffix.
What thinking level does Gemini 3.8 Flash use by default?+
Gemini 3.8 Flash defaults to medium; low and high are supported; minimal is not. Do not combine thinkingLevel with thinkingBudget.
Is Gemini 3.8 Flash always better than Gemini 3.1 Pro?+
No universal result is established here. Their defaults, release status and prices differ. Compare correctness, tool use and usage on representative tasks with explicit settings.
Can Gemini 3.8 Flash generate images or audio?+
Gemini 3.8 Flash analyzes supported media and returns text. Choose a dedicated model for image or audio generation.
Can I choose a Gemini 3.8 Flash account group?+
Select a Gemini 3.8 Flash group in the Playground, or send X-SeedRouter-Group in the API. Omit the header for automatic selection.
Try Gemini 3.8 Flash
Bring a prompt with a result you can check. Review the request, choose a group, and evaluate the response on your own task.