Gemini 3.8 Flash Is Now Available in Buda
Gemini 3.8 Flash brings long-horizon Agent work to every Buda plan at a displayed 0.2x Credit multiplier.
Gemini 3.8 Flash Is Now Available in Buda
Gemini 3.8 Flash is now available in Buda for all plans at a 0.2x Credit multiplier.
Google built this release for long-horizon software engineering, autonomous agents, and complex enterprise workflows. It keeps the Flash profile—fast and economical—but spends more reasoning steps and tool calls when a difficult task needs them.
That makes it a useful middle route: more capable than an efficiency-first model on work that has to plan, use tools, and recover, without sending every step to a frontier-priced model.
What changed in Gemini 3.8 Flash?
Google describes 3.8 Flash as its most intelligent Flash model. The release focuses on three areas:
- Long-horizon coding: stronger performance on engineering tasks that require multiple changes, tool calls, tests, and recovery.
- Agent work: better results in tool use and real-world knowledge-work evaluations.
- Specialized reasoning: more diligence on finance, legal, STEM, and other multistep professional tasks.
The model has a 1,048,576-token input limit and a 65,536-token output limit. It accepts text, images, video, audio, and PDFs, and returns text. Its Gemini API capabilities include caching, code execution, function calling, file search, search grounding, structured output, URL context, and preview computer use.
This is not an image, audio, or video generation model. Those jobs still belong on dedicated media models.
Gemini 3.8 Flash in Buda
Gemini 3.8 Flash is an independent option in Buda's model selector and Model API. It is active on every plan, including Free, with a displayed 0.2x Credit multiplier.
Buda does not expose a manual low, medium, or high thinking control for this model. The model handles its internal reasoning within the task instead.
Pricing and Credits are related, but not identical
Google's introductory API price through December 31, 2026 is:
- $0.75 per million input tokens
- $3.75 per million output tokens, including thinking tokens
Starting January 1, 2027, Google lists $1.50 input and $7.50 output per million tokens.
In Buda, Gemini 3.8 Flash is displayed at a 0.2x Credit multiplier. Actual Credits depend on input, output, cache, tools, and minimum-charge rules. The multiplier is a relative guide, not a fixed price per task.
When should you choose Gemini 3.8 Flash?
Start with it when a task has all three properties:
- It needs planning, tools, or several dependent steps.
- Speed and cost still matter enough that a premium frontier route is hard to justify.
- The result can be checked against a concrete artifact, test, or acceptance rule.
Good candidates include:
- implementing and testing a contained code change;
- researching across websites, PDFs, screenshots, and files;
- turning source material into a structured report or spreadsheet;
- maintaining a recurring workflow that must recover from ordinary tool failures;
- inspecting a repository and producing a reviewable patch.
Keep a smaller model for extraction, tagging, formatting, and other deterministic steps. Use a stronger specialist only when 3.8 Flash misses the acceptance bar.
Gemini 3.8 Flash is not Gemini 3.8 Flash Cyber
Google launched two related products. The regular Gemini 3.8 Flash is the general model available through consumer, developer, and enterprise surfaces. Gemini 3.8 Flash Cyber has more permissive cybersecurity mitigations and is limited to trusted defenders through Google's Fairwind Program.
Buda's model entry refers to Gemini 3.8 Flash, not Flash Cyber. Do not assume that selecting the general model enables the restricted cyber variant or its access policy.
Use it as part of a visible Agent workflow
A capable model still needs context, tools, permissions, and review.
In Buda, you can choose Gemini 3.8 Flash in the model selector and let an Agent work across persistent files and available tools. Keep consequential actions behind confirmation and inspect the resulting document, code, table, or other artifact before it leaves the workspace.
The practical test is not whether a benchmark score rose. It is whether the model produces accepted work with fewer retries, less reviewer correction, and an appropriate Credit cost.
For a general routing method, read How to Choose the Right Model for Your AI Agents. Current model multipliers are listed on Buda Pricing, and workspace controls are covered in the Buda documentation.
FAQ
Is Gemini 3.8 Flash available on Buda's Free plan?
Yes. It is available across Buda plans, including Free, at a displayed 0.2x Credit multiplier.
Can I set a thinking level in Buda?
No. Buda does not expose a manual thinking-level control for Gemini 3.8 Flash.
Does 0.2x mean every request costs the same?
No. It is a relative multiplier. Actual Credits depend on usage and tools.
Does Gemini 3.8 Flash generate images or video?
No. It can understand multimodal inputs, but its output is text. Use dedicated image or video models for media generation.
Is Flash Cyber also available in Buda?
No. The Buda entry is the regular Gemini 3.8 Flash. Google's Flash Cyber variant has a separate restricted-access program.