Can I run TranslateGemma 4B IT locally?
TranslateGemma 4B IT by google has 4.3B parameters. At 4-bit with an 8K context it needs about 3.5 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 8 GB of VRAM.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from bullerwins/translategemma-4b-it-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx | 128K ctx |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | rough | 2.6 GB | 2.7 GB | 3.1 GB | 5.0 GB |
| 3-bit (Q3_K_M) | usable | 3.0 GB | 3.1 GB | 3.6 GB | 5.5 GB |
| 4-bit (Q4_K_M) • | good, most popular | 3.4 GB | 3.5 GB | 4.0 GB | 5.8 GB |
| 5-bit (Q5_K_M) • | very good | 3.7 GB | 3.8 GB | 4.3 GB | 6.1 GB |
| 6-bit (Q6_K) • | near lossless | 4.1 GB | 4.1 GB | 4.6 GB | 6.5 GB |
| 8-bit (Q8_0) • | effectively lossless | 4.9 GB | 5.0 GB | 5.5 GB | 7.4 GB |
| 16-bit (F16/BF16) | original weights | 9.1 GB | 9.2 GB | 9.6 GB | 11.5 GB |
Which GPUs and Macs run TranslateGemma 4B IT?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
| Hardware | Memory | Fits? | Best quant | Needs | Max tok/s (est.) | Details |
|---|---|---|---|---|---|---|
| RTX 3060 12GB | 12 GB | Fits | F16 | 9.2 GB | 42 | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | F16 | 9.2 GB | 33 | details |
| RTX 3090 | 24 GB | Fits | F16 | 9.2 GB | 109 | details |
| RTX 4090 | 24 GB | Fits | F16 | 9.2 GB | 117 | details |
| RTX 5090 | 32 GB | Fits | F16 | 9.2 GB | 208 | details |
| RTX 5070 Ti | 16 GB | Fits | F16 | 9.2 GB | 104 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 9.2 GB | 112 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 9.2 GB | 32 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 9.2 GB | 63 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 9.2 GB | 95 | details |
| DGX Spark | 128 GB | Fits | F16 | 9.2 GB | 32 | details |
| RTX 4070 | 12 GB | Fits | F16 | 9.2 GB | 59 | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | F16 | 9.2 GB | 52 | details |
| RTX 4060 | 8 GB | Fits | Q8_0 | 5.0 GB | 66 | details |
| Arc B580 | 12 GB | Fits | F16 | 9.2 GB | 53 | details |
| RTX 3060 Ti | 8 GB | Fits | Q8_0 | 5.0 GB | 108 | details |
| RTX 3070 | 8 GB | Fits | Q8_0 | 5.0 GB | 108 | details |
| RTX 3080 10GB | 10 GB | Fits | Q8_0 | 5.0 GB | 184 | details |
| RTX 3080 12GB | 12 GB | Fits | F16 | 9.2 GB | 106 | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 9.2 GB | 117 | details |
| RTX 4060 Ti 8GB | 8 GB | Fits | Q8_0 | 5.0 GB | 70 | details |
| RTX 4070 Super | 12 GB | Fits | F16 | 9.2 GB | 59 | details |
| RTX 4070 Ti | 12 GB | Fits | F16 | 9.2 GB | 59 | details |
| RTX 4070 Ti Super | 16 GB | Fits | F16 | 9.2 GB | 78 | details |
| RTX 4080 | 16 GB | Fits | F16 | 9.2 GB | 83 | details |
| RTX 4080 Super | 16 GB | Fits | F16 | 9.2 GB | 86 | details |
| RTX 5060 | 8 GB | Fits | Q8_0 | 5.0 GB | 108 | details |
| RTX 5070 | 12 GB | Fits | F16 | 9.2 GB | 78 | details |
| RTX 5080 | 16 GB | Fits | F16 | 9.2 GB | 112 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 9.2 GB | 109 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 9.2 GB | 117 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 9.2 GB | 208 | details |
| RTX A6000 | 48 GB | Fits | F16 | 9.2 GB | 89 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 9.2 GB | 112 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 9.2 GB | 208 | details |
| L40S | 48 GB | Fits | F16 | 9.2 GB | 100 | details |
| A100 40GB | 40 GB | Fits | F16 | 9.2 GB | 181 | details |
| A100 80GB | 80 GB | Fits | F16 | 9.2 GB | 237 | details |
| H100 PCIe | 80 GB | Fits | F16 | 9.2 GB | 233 | details |
| H100 SXM | 80 GB | Fits | F16 | 9.2 GB | 390 | details |
| H200 | 141 GB | Fits | F16 | 9.2 GB | 558 | details |
| RX 7600 | 8 GB | Fits | Q8_0 | 5.0 GB | 70 | details |
| RX 7700 XT | 12 GB | Fits | F16 | 9.2 GB | 50 | details |
| RX 7800 XT | 16 GB | Fits | F16 | 9.2 GB | 73 | details |
| RX 7900 GRE | 16 GB | Fits | F16 | 9.2 GB | 67 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 9.2 GB | 93 | details |
| RX 9070 | 16 GB | Fits | F16 | 9.2 GB | 74 | details |
| RX 9070 XT | 16 GB | Fits | F16 | 9.2 GB | 74 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 9.2 GB | 30 | details |
| Arc A770 16GB | 16 GB | Fits | F16 | 9.2 GB | 65 | details |
| Mac M1 16GB | 16 GB | Fits | F16 | 9.2 GB | 7.9 | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 9.2 GB | 12 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 9.2 GB | 12 | details |
| Mac M4 16GB | 16 GB | Fits | F16 | 9.2 GB | 14 | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 9.2 GB | 14 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 9.2 GB | 23 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 9.2 GB | 17 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 9.2 GB | 32 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 9.2 GB | 32 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 9.2 GB | 47 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 9.2 GB | 47 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 9.2 GB | 47 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 9.2 GB | 63 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 9.2 GB | 93 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 9.2 GB | 95 | details |
Questions
How much VRAM does TranslateGemma 4B IT need?
About 3.5 GB at 4-bit (Q4_K_M) with an 8K context: 2.3 GB of weights (real GGUF file size), 0.3 GB of KV cache and 0.9 GB of runtime overhead. At 8-bit it needs about 5.0 GB.
What is the smallest GPU that runs TranslateGemma 4B IT?
At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 (8 GB).
Can I run TranslateGemma 4B IT on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M1 16GB or larger, using the default macOS GPU memory limit.
Similar models
Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.
Gemma 3 4B IT
4.3B · 3.5 GB at 4-bitMedGemma 1.5 4B IT
4.3B · 3.5 GB at 4-bitMedGemma 4B IT
4.3B · 3.5 GB at 4-bitQwen3 VL 4B Instruct
4.4B · 4.1 GB at 4-bitPhi 3.5 Vision Instruct
4.1B · 5.8 GB at 4-bitSingVL S1 4B
4.5B · 3.6 GB at 4-bitEmbed this
Free badge and "Will it run on my GPU?" widget for model cards, READMEs and blogs. No cookies, no tracking. All options.
Copy a snippet
Markdown badge
[](https://runsonmy.com/model/translategemma-4b-it/) Widget (script)
<div class="runsonmy-widget" data-model="translategemma-4b-it"><a href="https://runsonmy.com/model/translategemma-4b-it/">Will TranslateGemma 4B IT run on my GPU? Check on RunsOnMy</a></div>
<script async src="https://runsonmy.com/embed.js"></script> Widget (iframe)
<iframe src="https://runsonmy.com/embed/m/translategemma-4b-it/" width="100%" height="220" style="max-width:440px;border:0" loading="lazy" title="Will TranslateGemma 4B IT run on my GPU? (RunsOnMy)"></iframe> Model facts
| Hugging Face repo | google/translategemma-4b-it |
| Parameters | 4,300,079,472 (4.3B) |
| Layers / hidden size | 34 / 2560 |
| Attention | GQA + sliding window; 8 heads, 4 KV heads, head dim 256; sliding window 1024 |
| KV cache per token (FP16) | 136.0 KiB (sliding layers capped at the window) |
| Max context | 131,072 tokens |
| License | gemma (gated: accept terms on Hugging Face) |
| Released on HF | Jan 12, 2026 |
| Downloads (30d) / likes | 10,470 / 916 |
Sources: Hugging Face model API · config.json (mlx-community/translategemma-4b-it-4bit) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.