Gemma 3 4B
A smaller multimodal model to explore when your memory budget comes first.
- Published size
- ~4B
- Active parameters
- ~4B
- Context window
- 128K
- Architecture
- Dense · vision
- License
- Gemma
Deployment considerations
A smaller weight footprint leaves more room for serving overhead. It does not establish a particular latency or quality level.
Memory figures are calculated from the nominal 4B model size and exclude all serving overhead.
The instruction-tuned model accepts text and images. Measure the modalities you plan to use.
The 128K context capability still requires appropriate runtime configuration and memory. Review the Gemma license before deployment.
Give your model
some breathing room.
This is not a fit guarantee. KV cache, activations, quantization metadata, and the runtime still need memory.
Decimal GB; nominal parameter counts where marked ~. Bit widths illustrate weight storage, not validated quantizations. Understand the estimate
No invented leaderboards.
Latency, throughput, and cost per token will appear here after a reproducible run. Until then, this page helps you understand the model—not predict its performance.
Gemma 3 4B deployment FAQ
How much GPU memory does Gemma 3 4B need?
At 16-bit precision, the estimated weight storage is 8 GB. At 8-bit it is 4 GB, and at 4-bit it is 2 GB. These are theoretical weight-only estimates, excluding KV cache, runtime allocations, and quantization metadata. A working deployment needs additional memory and a supported checkpoint.
Has BenchGrid benchmarked Gemma 3 4B?
Not yet. This profile contains publisher specifications and calculated weight-memory estimates. We do not currently publish measured latency, throughput, or cost per token for this model.
Where do these specifications come from?
The specifications are based on the official Google model card linked on this page. Memory estimates use the stated total parameter count, including inactive experts for MoE models. Nominal model sizes are labeled with ~.
Explore your compute options.
Check available hardware, quotas, and current pricing with the provider. These links are not verified deployments or performance recommendations.
