Google Gemma 4 31B
Google's flagship 31B dense model with 256K context and top-tier text and image understanding.
From Google, Gemma 4 31B is a cloud text model. It is multimodal: alongside a text prompt it accepts images, then replies with generated text. Its context window handles up to 262K input tokens and up to 33K output tokens. Generation can be tuned with system instructions, temperature, and top-p sampling. It runs through Google using your own API key, from $0.00 per million input tokens. It also runs on your own hardware through Ollama, with no API key and no per-use fees.
- Released
- Apr 2026
- Pricing
- $0.00 / 1M in · $0.00 / 1M out
- Inputs
- Text, Images
- Context window
- 262K in · 33K out
- Controls
- System prompt, Temperature, Top-p
- Text rank
- #77
- Code rank
- #114
- Vision rank
- #45
Samples
Generated with Gemma 4 31B via Ollama. 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 translator that allows two different software programs to communicate. It lets one application request specific data or actions from another without needing to know how the other is built, ensuring different systems can work together seamlessly and securely.
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.
Use a hosted provider. For a two-person team, spending precious engineering hours on infrastructure like authentication is a mistake when you should be focusing on your core value proposition and achieving product-market fit. Hosted solutions handle security, encryption, and compliance out of the box, reducing your operational risk and time to launch. The strongest argument against this recommendation is vendor lock-in, as migrating your entire user database to a different system or your own internal solution later can be technically difficult and expensive as you scale.
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.
Predicting your company's AI expenditures is challenging because traditional software costs are static, while AI is dynamic. First, most services use token-based pricing, meaning costs fluctuate based on the volume of text processed rather than a flat monthly fee. Second, the field evolves rapidly, often forcing companies to upgrade to newer, more expensive models to remain competitive. Third, scaling AI introduces hidden infrastructure costs as data processing needs grow unpredictably with user adoption. These variables create a financial landscape where a small increase in usage can lead to a massive spike in bills. To manage this volatility, you should implement strict API usage caps and real-time monitoring to catch cost spikes before they escalate. Additionally, consider negotiating fixed-price contracts with vendors for predictable workloads. By shifting from a passive spending model to an active governance framework, you can transition from reactive surprises to strategic investments. AI is simply not a standard corporate line item; it is a utility that requires constant tuning. If you treat it like traditional software, your budget will inevitably drift. Instead, focus on visibility and constraints to ensure that innovation does not come at the expense of financial stability or operational predictability across the entire organization in the coming fiscal years.
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 ↗Google's flagship reasoning model for complex software engineering, deep mathematical and scientific problem solving and autonomous multi-step agents, with a shared 1M-token context.
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.
Google's most efficient multimodal Flash model: an updated reasoning stack with stronger coding and computer-use quality, at the same speed and scale as the rest of the Flash line.
Google's most capable multimodal model with deep reasoning and 1M token context.
Google's 26B mixture-of-experts model with long context and strong multimodal reasoning.
Google's updated Nano Banana 2 image model, with sharper graphic composition, steadier subjects across edits and closer prompt following, from text or up to 14 reference images at 1K to 4K.
Google's strongest open models dropped the custom license for plain Apache 2.0. Five sizes, one command each, phone to workstation.
Type “gemma” into your runtime and you get a wall of models: a phone-sized one, a workstation one, and odd cousins. Here’s which one you actually want.
Open weights now match GPT-4 quality. Here's how CSuite runs them on your machine: no proxy, no logging, no tokens billed.