Gemma 4 31B

Gemma 4 31B is part of the Gemma 4 family of open models built by Google DeepMind, released with open weights under the Apache 2.0 license in both pre-trained and instruction-tuned variants. The family is described as the most intelligent open models from DeepMind, built from Gemini 3 research and technology with a focus on maximizing intelligence per parameter. The 31B variant sits alongside four other sizes, giving it a position aimed at laptop and server deployment rather than the smaller E2B and E4B variants that target phones and IoT devices.

The model is multimodal, accepting text and image inputs while generating text output, and supports a context window of up to 256K tokens with multilingual coverage across more than 140 languages. The Gemma 4 lineup mixes Dense and Mixture-of-Experts architectures, and the 31B variant is presented as a dense configuration suitable for text generation, coding, and reasoning workloads. An instruction-tuned version, referenced as gemma-4-31b-it, is available through Google AI Studio, making it a practical choice for developers who want a capable open-weight model with strong reasoning behavior and flexible deployment on higher-end hardware.