GaMS3-12B-Instruct
GaMS3-12B-Instruct represents the next generation of the GaMS (Generative Model for Slovene) model. The model is based on Google's Gemma 3 family and continually pretrained on Slovene, English, and some portion of Croatian, Serbian, and Bosnian corpora. The supervised fine-tuning phase was done on a combination of Slovene and English datasets.
Params
11.8 B
Context
131,072
Downloads 30d
226 K
Likes
8
Download history
daily snapshots · 10 days226 K143 K
Nov 1Nov 4Nov 7Nov 10
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| model.safetensors | bf16 | 23.5 GB | 28.2 GB | ❌ Won’t fit |
| model.safetensors (bf16, full) | bf16 + 128K ctx | 23.5 GB | 54.6 GB | ❌ Won’t fit |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Run it
copy-paste, exact tags checked against the Hub$ curl -s https://aimodelscomparison.com/api/v1/models/gams3-12b-instruct
{
"hf_id": "cjvt/GaMS3-12B-Instruct",
"params_b": 11.77,
"context_length": 131072,
"license": { "id": "gemma", "commercial": "conditional" },
"downloads_30d": 225941,
"vram_estimates": [
{ "quant": "bf16", "gb": 28.2 }
],
"updated_at": "2026-11-01T01:00:19Z"
}
Specifications
- Architecture
- Gemma3ForCausalLM
- Parameters
- 11.8 B
- Tensor type
- BF16
- Context length
- 131,072
- Vocabulary
- 262,208
- Layers / heads
- 48 / 16
- Licence
- gemma
- First seen on the Hub
- 2025-12-04
- Training datasets
- undisclosed
- Added to our catalog
- 2026-11-01
Compare with any text-generation model
Popular comparisons
Appears in