Gemma 3 1B VRAM Requirements
Gemma 3 1B needs about 1.6 GB of VRAM at Q4_K_M with an 8K context: 0.6 GB for weights, 43 MB for the KV cache and 1.0 GB of runtime overhead. At Q8_0 it needs 2.0 GB and at full FP16 precision 2.9 GB. The smallest GPU that runs it fully is the RTX 3050 8GB.
1.6 GB VRAM
Gemma 3 1B at Q4_K_M with 8K context and the KV cache in FP16.
- Weights 0.6 GB
- KV cache 43 MB
- Runtime overhead 1.0 GB
Which GPUs can run it
- RTX 3060 12GBFits~336 tok/s
- RTX 4060 Ti 16GBFits~274 tok/s
- RTX 3090 24GBFits~767 tok/s
- Radeon RX 7900 XTX 24GBFits~684 tok/s
- RTX 4090 24GBFits~814 tok/s
- RTX 5090 32GBFits~1243 tok/s
- RTX 6000 Ada 48GBFits~783 tok/s
- A100 80GBFits~1354 tok/s
- H100 80GB SXMFits~1814 tok/s
- Ryzen AI Max+ 395 128GBFits~228 tok/s
- Apple M4 Max 128GBFits~488 tok/s
- Apple M5 Max 128GBFits~541 tok/s
- RTX PRO 6000 Blackwell 96GBFits~1243 tok/s
- NVIDIA DGX Spark 128GBFits~242 tok/s
- Apple M3 Ultra 512GBFits~689 tok/s
Gemma 3 1B VRAM by quantization
Total memory with an FP16 KV cache. "Smallest tier" is the smallest standard VRAM size with a 6% safety margin at 8K.
| Quantization | Bits/weight | Weights | Total at 8K | Total at 32K | Smallest tier |
|---|---|---|---|---|---|
| Q4_K_M | 4.85 | 0.6 GB | 1.6 GB | 1.7 GB | 4 GB |
| Q5_K_M | 5.69 | 0.7 GB | 1.7 GB | 1.8 GB | 4 GB |
| Q6_K | 6.59 | 0.8 GB | 1.8 GB | 1.9 GB | 4 GB |
| Q8_0 | 8.50 | 1.0 GB | 2.0 GB | 2.1 GB | 4 GB |
| FP16 / BF16 | 16.00 | 1.9 GB | 2.9 GB | 3.0 GB | 4 GB |
| Q3_K_M | 3.91 | 0.5 GB | 1.5 GB | 1.6 GB | 4 GB |
| Q2_K | 3.00 | 0.3 GB | 1.4 GB | 1.5 GB | 4 GB |
VRAM by context length
At Q4_K_M. 4 of 26 layers keep the full context. The other 22 use a 512-token sliding window, so their cache stops growing after 512 tokens.
| Context | KV cache (FP16) | KV cache (Q8_0) | Total, FP16 cache | Total, Q8_0 cache |
|---|---|---|---|---|
| 4K tokens | 27 MB | 14 MB | 1.6 GB | 1.6 GB |
| 8K tokens | 43 MB | 23 MB | 1.6 GB | 1.6 GB |
| 32K tokens | 0.1 GB | 74 MB | 1.7 GB | 1.7 GB |
GPUs that can run Gemma 3 1B
At Q4_K_M with 8K context. "Best quant" is the highest-quality format that fits fully; "max context" is at Q4_K_M.
| GPU | Usable memory | Status | Best quant | Max context | Speed |
|---|---|---|---|---|---|
| RTX 3050 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~217 tok/s |
| RTX 4060 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~260 tok/s |
| RTX 4060 Ti 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~274 tok/s |
| Radeon RX 7600 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~234 tok/s |
| RTX 3060 Ti 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~410 tok/s |
| RTX 3070 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~410 tok/s |
| RTX 5060 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~410 tok/s |
| RTX 5060 Ti 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~410 tok/s |
| RTX 3070 Ti 8GB | 8.0 GB | Fits | FP16 / BF16 | 32K | ~536 tok/s |
| Intel Arc B570 10GB | 10.0 GB | Fits | FP16 / BF16 | 32K | ~278 tok/s |
| RTX 3080 10GB | 10.0 GB | Fits | FP16 / BF16 | 32K | ~647 tok/s |
| Apple M4 16GB | 10.7 GB | Fits | FP16 / BF16 | 32K | ~119 tok/s |
| RTX 2080 Ti 11GB | 11.0 GB | Fits | FP16 / BF16 | 32K | ~542 tok/s |
| RTX 2060 12GB | 12.0 GB | Fits | FP16 / BF16 | 32K | ~316 tok/s |
About Gemma 3 1B
Google's smallest Gemma 3, text only. It uses a single KV head and sliding-window attention on most layers, so its context memory is tiny. The 262K vocabulary takes up about a third of its parameters.
| Spec | Value |
|---|---|
| Parameters | 1B |
| Layers | 26 |
| Hidden size | 1152 |
| Attention heads / KV heads | 4 / 1 |
| Head dimension | 256 |
| Vocabulary | 262,144 |
| Max context (used here) | 32K tokens |
| Sliding window | 512 tokens on 22 of 26 layers |
| KV cache per token (FP16) | 4 KB |
| Released | 2025-03 |
| License | Gemma Terms of Use |
Gemma 3 1B VRAM: common questions
How much RAM does Gemma 3 1B need?
Under 2 GB at Q4_K_M even with its full 32K context, so it runs on almost any device.
How much VRAM does Gemma 3 1B need?
About 1.6 GB at Q4_K_M with an 8K context, 2.0 GB at Q8_0 and 2.9 GB at FP16. Longer contexts add 0.00 GB per 1,000 tokens with an FP16 cache.
What is the cheapest GPU that can run Gemma 3 1B?
The smallest listed device that fits it fully at Q4_K_M with 8K context is the RTX 3050 8GB (8.0 GB usable).