Gemma 4 31B IT

Gemma 4 31B IT is Google's instruction-tuned, 31-billion-parameter dense model. It is available through the Gemini API for text generation, coding, reasoning, and image understanding.

Capabilities and context limits

Gemma 4 31B IT accepts text and image inputs and generates text responses. Its dense architecture uses approximately 30.7 billion parameters, with long-context support for document analysis and conversations. Google's Gemini API model listing reports an input limit of 262,144 tokens and an output limit of 32,768 tokens.

The model can follow instructions, work with code, and reason about images alongside text. As with other language models, answer quality and latency depend on prompt design, input length, and task complexity. It does not provide native audio input or audio output.

API access and pricing

The Gemini API identifies this model as gemma-4-31b-it. Google's Gemma API guide demonstrates the generateContent endpoint for the Gemma 4 family. Developers can select this Google model in EidoStack, supply their own Gemini API key, and inspect text responses and token usage.

Google's Gemini API pricing page lists free-tier input and output tokens for Gemma 4 and no published paid-tier token rate. Provider quotas and availability still apply; free pricing does not guarantee unrestricted requests.

When to use this model

The 31B dense variant is a useful candidate for complex text analysis, coding assistance, and image-based reasoning where model quality matters more than minimum compute cost. For latency-sensitive comparisons, consider testing the mixture-of-experts Gemma 4 26B A4B IT against the same prompts before choosing a model.

For primary specifications, see Google's Gemma 4 model card. The API-specific input and output limits above are from Gemini API model-list metadata published in the Google AI Developers Forum.