For Developers/Models/Gemini 3.8 Flash
Closed SourceGoogleReleased 2026-09

Gemini 3.8 Flash

Most intelligent workhorse model with enhanced reasoning and coding

Context window

1M

Input / 1M tokens

$0.75

Output / 1M tokens

$3.75

Provider

Google

Data verified 2026-09-03

Gemini 3.8 Flash delivers significant improvements over 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning. It performs additional reasoning steps and iterative tool calling on complex tasks. The model maintains a 1M token context window with multimodal input support.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding
8
Reasoning
8
Math
7
Multimodal
8
Long context
8
Speed
8
Cost efficiency
8

How to access it

Available to Google AI Pro and Ultra subscribers via Gemini app, AI Mode in Google Search, Gemini in Google Sheets. Developers can access via Google AI Studio, Gemini API, and Google Antigravity.

Strengths

  • Long-horizon software engineering
  • Agentic automation and workflows
  • Multi-step reasoning in specialized domains
  • Cost-efficient frontier-level performance

Best for developers who...

Software engineering tasksAutonomous agentsComplex enterprise analysisFinancial and legal workflows

When to choose it (and when not to)

Reach for Gemini 3.8 Flash when...

  • When better completion rates justify higher token consumption
  • For complex, multi-step reasoning tasks
  • Agentic workflows requiring iterative tool use

Look elsewhere if...

  • Latency-sensitive applications
  • High-concurrency pipelines
  • Tightly budgeted inference workloads

How to use it

  • Leverage effort levels to trade quality vs cost/latency
  • Use for agent-first workflows
  • Suitable for long-running agentic loops

Quickstart

Python
import anthropic

client = anthropic.Anthropic(api_key="your-google-api-key")
response = client.messages.create(
  model="gemini-3.8-flash",
  messages=[{"role": "user", "content": "Hello"}]
)
print(response.content)

Access via Google AI Studio or Gemini API

API model id: gemini-3.8-flash

Benchmarks

BenchmarkScoreNotes
DeepSWE v1.1Outperforms most larger frontier modelsLong-horizon software engineering benchmark
HLE-Verified54.9%Multi-step reasoning across STEM, humanities, and professional fields
Vals Finance Agent V2Beats 3.7 Flash and frontier modelsFinancial analysis benchmark
Harvey Legal Agent BenchmarkTops previous iterationsLegal domain benchmark

Source: Google Official Announcement

Compare Gemini 3.8 Flash

Compare Gemini 3.8 Flash with any other model

Build a comparison →
All model comparisons →

Learn the concepts