Google Gemini 3.8 Flash
Google's most capable Flash model, with clear gains over 3.7 Flash in software engineering, agentic tasks, and multi-step reasoning, plus vision and a 1M-token context window.
Gemini 3.8 Flash is a cloud text model built by Google. It is multimodal: alongside a text prompt it accepts images, then replies with generated text. Its context window handles up to 1.05M input tokens and up to 66K output tokens, and it supports adjustable reasoning effort for harder problems. Generation can be tuned with system instructions, temperature, and top-p sampling. It runs through Runware, OpenRouter, and Google using your own API key, from $0.75 per million input tokens.
- Released
- this month (Sep 2026)
- Pricing
- $0.75 / 1M in · $3.75 / 1M out
- Inputs
- Text, Images
- Context window
- 1.05M in · 66K out
- Reasoning
- Adjustable effort
- Controls
- System prompt, Temperature, Top-p
Examples
Generated with Gemini 3.8 Flash via Runware. The same three prompts run against every text model in the catalog, shown verbatim, so the only thing that changes between two models’ answers is the model.
Explain what an API is to someone who has never written code. Do not use a restaurant or waiter analogy. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 50 words.
An API is a digital bridge that allows different software applications to talk to each other. For example, when you check the weather on your phone, the app uses an API to request the forecast from a distant weather station's computer system and display it on your screen.
A two-person startup is deciding whether to build their own user authentication or use a hosted provider. Give them a clear recommendation, then the single strongest argument against your own recommendation. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 150 words.
You should use a hosted authentication provider. With only two people, your immediate survival hinges entirely on finding product-market fit, and building authentication is undifferentiated, time-consuming work. Rolling your own system introduces dangerous security liabilities, compliance headaches, and weeks of distraction away from the unique features that actually validate your business. Managed services allow you to ship in hours with robust security, session management, and social logins already solved. The strongest argument against this recommendation is the risk of painful vendor lock-in combined with steep scaling costs. Hosted identity platforms frequently charge aggressive per-user pricing that balloons rapidly as your user base grows. Migrating thousands of active users, hashed passwords, and existing sessions away from a proprietary system later is an engineering nightmare that could stall your product momentum right when you finally achieve traction.
Write a briefing for a non-technical executive explaining why their company's spending on AI is hard to predict, and what they can do about it. Cover at least three distinct causes. Plain prose only — no headings, bullet points, or markdown formatting. Write exactly 205 words.
Forecasting artificial intelligence spending poses a significant challenge for executive leadership because modern tools operate under fundamentally variable cost structures. First, vendors charge based on consumption rather than fixed licenses, meaning expenses fluctuate directly with user engagement, query complexity, and the volume of processed text. A sudden surge in customer adoption can quickly inflate your monthly invoice. Second, development is inherently experimental. Technical teams frequently test multiple competing models, run thousands of trial queries, and reprocess data to improve accuracy, making the research phase exceptionally difficult to budget reliably. Third, hidden operational costs compound silently through data storage, system integration, security monitoring, and unexpected vendor pricing updates. To regain financial control, your organization must implement active governance rather than passive monitoring. You should establish automated spending caps and immediate budget alerts across all vendor accounts to prevent runaway expenses. Next, form a centralized review committee to evaluate proposed projects, requiring business units to prove measurable business value before scaling expensive computational resources. Finally, mandate strict architectural discipline by encouraging technical teams to deploy smaller, specialized models instead of massive foundation models for basic corporate workflows. Through clear institutional guardrails, regular financial reviews, and prudent architectural choices, you can foster competitive innovation while maintaining fiscal predictability.
Google and Google DeepMind build the Gemini family of multimodal models, the Imagen and Nano Banana image models, the Lyria music models, and the Veo video models.
deepmind.google ↗