CrossModel
Gemini 3.8 Flash
gemini/gemini-3.8-flash
Back to model catalog

Gemini · Model guide

Gemini 3.8 Flash

Compare
gemini/gemini-3.8-flashReleased Sep 2, 202650% off

Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents and complex enterprise workflows. It keeps the 1M-token multimodal context, 64K output and built-in tool suite of the Flash line, with three thinking levels (low, medium, high). By design it spends more tokens on long multi-step tasks — taking smaller reasoning steps, calling tools iteratively and verifying its work — so lower the thinking level for everyday requests.

Modalities
TextImageAudioVideoText
Context
1M
Max output
66K
Price / 1M tokens
$ 1.50$ 0.75in$ 3.75out
StreamingTool callingJSON outputImage inputThinking mode

What this model is good at

What the vendor positions it for, and which of our models to reach for instead.

VendorGoogle's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows.source

Best for

  • Long-horizon software engineering · Google reports strong results on real-world coding benchmarks, complex multi-file refactoring and deterministic tool execution.(vendor claim)source

  • Complex enterprise workflows · Google positions it for demanding domain tasks and large-scale data pipelines requiring deep reasoning and factual rigor.(vendor claim)source

  • Long multimodal workflows · It accepts text, images, audio, video and PDFs in a 1M-token window, with context caching and built-in tools.source

Neighbouring models
  • gemini/gemini-3.7-flashpredecessor —Choose 3.7 Flash for everyday coding and tool use with fewer tokens spent on verification; Google states it remains fully supported.
  • gemini/gemini-3.5-flash-litecheaper —Choose 3.5 Flash-Lite when minimum latency and cost dominate.
Compare them side by side →

Pricing and billing

Billed per token. Cached input is charged at the cache-read rate.

Price / 1M tokensYou pay
Input$ 1.50$ 0.75
Output$ 7.50$ 3.75
Cache read$ 0.15$ 0.075
Cache write$ 1.50$ 0.75
Estimate a request
$ 0.0150
Estimated cost per request
10.0K × $ 0.75
+ 2.0K × $ 3.75
≈ $ 15.00 per 1,000 requests

Capabilities and limits

What CrossModel guarantees across every route this model can take right now.

Context window
1.0M tokens
Max output
65.5K tokens
Input / output modalities
Text + Image + Audio + Video → Text
Streaming
Supported
Tool calling
Supported
Structured output (JSON)
Supported
Image input
Supported
Thinking mode
Supported· always on
Reasoning effort
low · medium · high
Available endpoints
/v1/chat/completions · /v1/responses · /v1/messages
First-party sources:Specsai.google.dev 1deepmind.google 2ai.google.dev 3Capabilitiesai.google.dev 1ai.google.dev 2

Published benchmarks

Scores the vendor reported, with the evaluation setup each one came from.

Agentic coding
  • Terminal-Bench 2.1
    pass_rate·Google DeepMind · 2026-09-02setup not fully disclosed

    Agentic terminal coding. The 3.7 Flash column reads 85.8%, matching the value stored for that model.

    89.4%
  • Terminal-Bench 4.0
    pass_rate·Google DeepMind · 2026-09-02setup not fully disclosed

    General agent capabilities; the hardest row in Google's table and far below its Terminal-Bench 2.1 score.

    19.1%
Coding
  • DeepSWE 1.1
    resolved_rate·Google DeepMind · 2026-09-02setup not fully disclosed

    Long-horizon software engineering. Same table reports Gemini 3.7 Flash at 65.3%.

    73.7%
Computer use
  • OSWorld 2.0 (partial credit)
    success_rate·Google DeepMind · 2026-09-02setup not fully disclosed

    Partial score with the batch tool enabled. Kept separate from the plain OSWorld 2.0 rows in this catalog: this table reports 3.7 Flash at 50.6% under the same protocol, while the stored 3.7 Flash OSWorld 2.0 row is 47.9% from its own model card.

    59%
Knowledge work
  • Finance Agent v2
    score·Google DeepMind · 2026-09-02setup not fully disclosed

    Vals AI Finance Agent v2, financial analyst tasks.

    61.4%
  • GDPval-AA v2
    elo·Google DeepMind · 2026-09-02setup not fully disclosed

    Knowledge work. This table reports Gemini 3.7 Flash at 1482 Elo, while the 3.7 Flash model card reported 1525; both are kept under their own source rather than reconciled.

    1545
Multimodal reasoning
  • CharXiv Reasoning (no tools)
    accuracy·Google DeepMind · 2026-09-02setup not fully disclosed

    Information synthesis from complex charts, no tools.

    86.2%
Reasoning
  • HLE-Verified
    accuracy·Google DeepMind · 2026-09-02setup not fully disclosed

    Multidisciplinary expert reasoning.

    54.9%
Video understanding
  • LVBench (agentic)
    accuracy·Google DeepMind · 2026-09-02setup not fully disclosed

    Long video understanding with agentic navigation of the video timeline.

    87.8%
  • LVBench (static)
    accuracy·Google DeepMind · 2026-09-02setup not fully disclosed

    Long video understanding with static frame processing; the same table reports only one figure (85.4%) for Gemini 3.7 Flash.

    87.1%

Vendor-reported numbers, not CrossModel measurements. Scores are only comparable when the benchmark version, metric and evaluation setup match, so nothing here is averaged or ranked.

Use it in your tools

Point the base URL at CrossModel and paste this model ID — every tool below has a setup guide.

Claude Code
Anthropic's official command-line coding assistant
Cursor
AI-first code editor
Cline
Open-source AI coding agent for VS Code
Codex
OpenAI's coding agent for ChatGPT desktop and the terminal
OpenCode
Open-source terminal coding assistant
Zed
High-performance, AI-native code editor
Open WebUI
Open-source, self-hosted AI chat interface
Continue
Open-source AI coding assistant for VS Code & JetBrains
See all integration guides →

Frequently asked questions

What is Gemini 3.8 Flash?
Gemini 3.8 Flash is available on CrossModel as gemini/gemini-3.8-flash. It has a 1M-token context window and can return up to 66K tokens per request.
How much does Gemini 3.8 Flash cost?
Currently $ 0.75 per 1M input tokens and $ 3.75 per 1M output tokens. Cached input is billed at $ 0.075 per 1M.
Does Gemini 3.8 Flash support tool calling and structured output?
Tool calling: Supported · Structured output (JSON): Supported · Image input: Supported · Streaming: Supported · Thinking mode: Supported
Which endpoint do I call?
For text models, OpenAI Chat Completions (/v1/chat/completions), OpenAI Responses (/v1/responses) and Anthropic Messages (/v1/messages) all work — pick whichever your SDK already speaks.
Can I try it without writing code?
Yes. Sign in and open the Playground on this page to chat with the model directly — no API key needed. Usage is billed from your wallet at the normal rate.

Chat first, integrate later

Chat first, then wire it in once you like the answers.