Gemma 4 31B — Multimodal Language Models AI Model
Gemma 4 31B is Google DeepMind's flagship open vision-language model: a prompt plus up to two reference images in, text out. Visual Q&A, captioning, comparison, 256K context and an optional thinking mode.
Also runs in One Answer, Checked by Three AIs · Outfit Flat-Lay Collage
What it does
Gemma 4 31B is Google DeepMind's flagship open-weight model and the most capable of the Gemma 4 family.
It is multimodal and instruction-tuned: it reads text, images and video together, handles documents and code, works through multi-step reasoning, and can call tools to act on a task rather than only describe it.
When to use it
- Reasoning tasks where an open model has to stand in for a closed one
- Reading text, images and video together in a single conversation
- Working through a long document, report or codebase in one pass
- Generating, debugging and reviewing code at a competitive standard
- Driving tools and APIs autonomously across a multi-step task
Details
- CategoryMultimodal Language Models
- Year2026
- LicenseApache-2.0
Compliance & Provenance
- ProviderGoogle (open weights) · GPAI
- EU AI Act RiskLimited Risk
- Art. 50 TransparencyRequired — AI-generated outputs are marked
Inputs & Outputs
- TextInputstring
Text prompt, question, or instruction for the model (e.g. 'Describe this image in detail', 'What objects are visible?', 'Summarize the scene'). Used standalone for pure text generation or combined with an image for visual understanding tasks.
- Image 1Inputimageoptional
Optional reference image for visual understanding. When provided, the model analyzes the image content together with the text prompt. Supports up to 4096x4096 resolution. Connect from any image-producing node (e.g. super-resolution, deblur, inpainting) to analyze processed results.
- Image 2Inputimageoptional
Optional second reference image. When provided alongside Image 1, the model can compare, contrast, or jointly reason over both images with the text prompt. Supports up to 4096x4096 resolution.
- TextOutputstring
Generated text response from the model. Content depends on the input: image description, answer to a visual question, or general text generation. Can be connected to downstream text-consuming nodes (e.g. Text Processing, conditional gates).
Parameters
- max_new_tokensintegerdefault 2048range 1–8192advanced
Maximum tokens to generate (1 - 8192)
- temperaturefloatdefault 1range 0–2advanced
Sampling temperature (0.0 - 2.0)
- top_pfloatdefault 0.95range 0–1advanced
Top-p (nucleus)
Nucleus sampling threshold — keeps the smallest token set whose cumulative probability ≥ top_p. 0.95 (default) is a strong default. Lower values produce more focused output.
- top_kintegerdefault 64range 1–100advanced
Top-k sampling (1 - 100)
- enable_thinkingbooleandefault Offoptions: On, Offadvanced
Enable thinking mode for chain-of-thought reasoning before answering
Tags
Alternatives in Multimodal Language Models
- Agnes-3.0-Flash
Agnes 3.0 Flash Preview — 33B hybrid-attention vision-language model, 262K context, bf16. Open-weights preview checkpoint, distinct from the production Agnes 3.0 Flash.
- Gemma4-E2B
Gemma 4 E2B compact VLM, 128K context, bf16.
- Mage-VL
Mage-VL — 4B codec-native streaming multimodal model from Microsoft. A from-scratch Mage-ViT visual encoder on a Qwen3-4B decoder, trained to read images and video through codec structure rather than uniform frame sampling. This entry serves its image-and-text mode; its video and proactive-streaming modes are not wired up here.
- Qwen3.6-35B-MoE
Qwen 3.6 35B/A3B MoE VLM, 256K context. fast inference.
- Qwen3.8-27B
Qwen 3.8 27B Dense VLM, 256K context, bf16.