Gemini 3.8 Flash 是 Google 专为高级编程、智能体工作流和复杂多步推理打造的 AI 模型。它具备强大的软件工程、工具调用和专业推理能力,同时保持了 Flash 系列的极速响应与极致性价比。

Model Type:
Pricing: Input 45 credits / 1M tokens (≈ $0.225), Output 225 credits / 1M tokens (≈ $1.125) — both ~15% of official pricing. Limited-Time Offer: 50% OFF until Dec 31, 2026 at 06:00 UTC (prices shown are discounted). High-tier top-ups include +10% bonus credits, lowering the effective price by about 9%.

Automatically append previous messages to maintain multi-turn context. May increase token usage.

Function calling
Grounding with Google Search
Source:Google Search
Thinking process
Stream Response

How can I help you today?

README

Complete guide to using gemini-3-8-flash

专为高级编程与 AI 智能体打造的 Gemini 3.8 Flash API

Original image
使用 Google Gemini 3.8 Flash API 实现高级编程
专为长周期智能体工作流打造的 Gemini 3.8 Flash
借助 Gemini 3.8 Flash API 实现迭代式工具调用
面向金融、法律和 STEM 领域的 Gemini 3.8 Flash 推理能力
借助 Google Gemini 3.8 Flash 灵活调整推理力度
面向生产级应用:高速、高性价比的 Gemini API
Gemini 3.8 Flash DeepSWE v1.1 编码基准测试
Gemini 3.8 Flash HLE-Verified 推理基准测试
针对金融与法律 AI 的 Gemini 3.8 基准测试
核心能力Gemini 3.8 FlashGemini 3.7 Flash
Typical WorkloadsLong-horizon coding, agentic tasks, complex reasoningCoding, agentic tasks, efficiency-focused workloads
DeepSWE v1.173.7%65.3%
Vals Finance Agent v261.4%59.0%
HLE-Verified54.9%53.6%
OSWorld-2.059.0%50.6%
Reasoning BehaviorMore reasoning steps on complex tasksLower reasoning overhead
Tool UseMore iterative tool callingEfficiency-oriented tool use