Spark-X2.5-4B-Heretic-jp-gguf
GGUF quants of a Japanese-adapted, refusal-attenuated derivative of the XHToken/Spark-X2.5-4B model.
Params
4.0 B
Context
—
Downloads 30d
206 K
Likes
17
Download history
tracking started — chart appears after 7 days of snapshots (4 recorded)206 K downloads in the last 30 days
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 |
|---|---|---|---|---|
| Spark-X2.5-4B-Heretic-jp-Q4_K_M.gguf | Q4_K_M | 2.6 GB | 4.0 GB | ✅ Runs comfortably |
| Spark-X2.5-4B-Heretic-jp-Q5_K_M.gguf | Q5_K_M | 3.0 GB | 4.4 GB | ✅ Runs comfortably |
| Spark-X2.5-4B-Heretic-jp-Q6_K.gguf | Q6_K | 3.4 GB | 4.8 GB | ✅ Runs comfortably |
| Spark-X2.5-4B-Heretic-jp-Q8_0.gguf | Q8_0 | 4.4 GB | 5.9 GB | ✅ Runs comfortably |
| Spark-X2.5-4B-Heretic-jp-BF16.gguf | GGUF | 8.2 GB | 10.2 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 spark-x2-5-4b-heretic-jp-gguf # pin the quantization explicitly $ ollama run spark-x2-5-4b-heretic-jp-gguf-q4_k_m
$ huggingface-cli download soyaakinohara/Spark-X2.5-4B-Heretic-jp-gguf-GGUF \
Spark-X2.5-4B-Heretic-jp-Q4_K_M.gguf --local-dir .
$ llama-cli -m Spark-X2.5-4B-Heretic-jp-Q4_K_M.gguf \
-c 8192 -ngl 99 -t 8 --color
$ curl -s https://aimodelscomparison.com/api/v1/models/spark-x2-5-4b-heretic-jp-gguf
{
"hf_id": "soyaakinohara/Spark-X2.5-4B-Heretic-jp-gguf",
"params_b": 4.00,
"context_length": null,
"license": { "id": "apache-2.0", "commercial": "yes" },
"downloads_30d": 206274,
"vram_estimates": [
{ "quant": "GGUF", "gb": 10.2 },
{ "quant": "Q4_K_M", "gb": 4.0 }
],
"updated_at": "2026-10-22T01:00:17Z"
}
Specifications
- Parameters
- 4.0 B
- Licence
- apache-2.0
- First seen on the Hub
- 2026-09-04
- Training datasets
- fn-aka-mur/wiki40b_ja, izumi-lab/llm-japanese-dataset-vanilla, nappa0326/glaive-function-calling-v2-sharegpt-japanese
- Added to our catalog
- 2026-10-22
Compare with any text-generation model