Gemini 3.8 Flash
GeminiBalanced
ThinkingTool UseVisionStructured Output
About this model
Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.
Performance Tier
Balanced
Gemini 3.8 Flash is a balanced model from Gemini : strong performance at a reasonable price.
Strong cost-performance ratio. Reliable for most professional use cases without premium pricing.
Pricing
This model is included in Elosia plansModerate
Moderate cost. A balanced choice for regular use.
| Type | per 1M tokens |
|---|---|
| Input (prompt) | $0.825 |
| Output (completion) | $4.13 |
| Image | $0.825 |
| Internal reasoning | $4.13 |
| Cache read | $0.083 |
| Cache write | $0.042 |
Capabilities
Context Length1.0M
Max Output Tokens66K
TokenizerGemini
Inputtext, image, video, file, audio
Outputtext
Release DateSeptember 2, 2026
Benchmarks
General Intelligence
MMLU
Not reported
GPQA Diamond
95.3%
Mathematics
MATH-500
Not reported
Programming
HumanEval
Not reported
Reasoning
Humanity's Last Exam
47.8%
Multimodal
MMMU-Pro
85.6%
Agentic
Terminal-Bench 2.1
89.4%
Where does Gemini 3.8 Flash stand?
Compare its performance index against every other model.
Recommended Use Cases
CodingAnalysisGeneral ChatResearchData Extraction
Strengths
- Built for production agents: runs extra reasoning steps and calls tools iteratively on complex tasks
- Adjustable effort levels to trade quality against cost and latency
- 1M-token context window with 64K output
- Natively multimodal input covering text, images, audio, video and PDFs, including long video understanding
- Multi-step reasoning in specialized domains such as finance and legal agent workflows
Limitations
- Google's Terminal-Bench 2.1 score is self-computed; independent runs measure it lower
- Google warns it may consume more tokens at higher effort levels, with occasional slowness or timeouts
- Knowledge is uneven: some domains stop well before the stated knowledge cutoff, and Google reports weaker safety performance outside English
- Google publishes no GPQA Diamond, MMMU-Pro or Humanity's Last Exam score; those shown come from third-party testing
Frequently asked questions
Resources
This model may use your data for training
