CAJAL-4B
python from transformers import AutoModelForCausalLM, AutoTokenizer model = AutoModelForCausalLM.frompretrained("Agnuxo/CAJAL-4B", torchdtype="auto", devicemap="auto") tokenizer = AutoTokenizer.frompretrained("Agnuxo/CAJAL-4B")
Params
—
Context
—
Downloads 30d
219 K
Likes
1
Download history
daily snapshots · 26 days291 K191 K
Aug 27Sep 4Sep 13Sep 21
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 |
|---|---|---|---|---|
| CAJAL-4B-q4_k_m.gguf | Q4_K_M | 2.7 GB | 3.5 GB | ✅ Runs comfortably |
| CAJAL-4B-q8_0.gguf | Q8_0 | 4.5 GB | 5.4 GB | ✅ Runs comfortably |
| CAJAL-4B-f16.gguf | GGUF | 8.4 GB | 9.8 GB | ✅ Runs comfortably |
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$ ollama run cajal:4b # pin the quantization explicitly $ ollama run cajal:4b-q4_k_m
$ huggingface-cli download Agnuxo/CAJAL-4B-GGUF \
CAJAL-4B-q4_k_m.gguf --local-dir .
$ llama-cli -m CAJAL-4B-q4_k_m.gguf \
-c 8192 -ngl 99 -t 8 --color
$ curl -s https://aimodelscomparison.com/api/v1/models/cajal-4b
{
"hf_id": "Agnuxo/CAJAL-4B",
"params_b": null,
"context_length": null,
"license": { "id": "apache-2.0", "commercial": "yes" },
"downloads_30d": 218640,
"vram_estimates": [
{ "quant": "GGUF", "gb": 9.8 },
{ "quant": "Q4_K_M", "gb": 3.5 }
],
"updated_at": "2026-08-27T01:00:36Z"
}
Specifications
- Licence
- apache-2.0
- First seen on the Hub
- 2026-05-07
- Training datasets
- undisclosed
- Added to our catalog
- 2026-08-27
Compare with any text-generation model