Gemma 4 12B

Gemma 4 12B is a text chat model released by Google as part of its Gemma line. The "12B" designation refers to its approximate parameter count, which places it in the mid-size band of the family — larger than the compact variants intended for on-device or very high-volume use, and smaller than the largest Gemma configurations. Gemma models are Google's separately branded line from Gemini, and are typically served across a range of third-party inference providers rather than a single first-party endpoint, which is why the pricing table on this page may show several hosts for the same model. On measured serving performance, third-party benchmarking from Artificial Analysis records Gemma 4 12B at approximately 141 output tokens per second with a time to first token of roughly 1.34 seconds. The throughput figure is characteristic of a model in this parameter range: fast enough for interactive chat and streaming responses, while the first-token latency indicates it is better suited to conversational and batch workloads than to latency-critical paths where sub-second response initiation is required. These numbers reflect a specific provider configuration and will vary meaningfully depending on which host you select, the hardware behind it, and the batch conditions at the time of the request. In practice, a 12B-class model is commonly chosen when a smaller model is not accurate enough for the task but a large frontier model is more capacity than the workload justifies. Typical deployments include assistant-style chat, summarization, drafting, classification with reasoning, and retrieval-augmented question answering. We do not currently track a confirmed context window, modality list, or benchmark accuracy scores for Gemma 4 12B in our database, so evaluate those specifics against your target provider's documentation before committing to a deployment.