JiRackDeltaNet_27b
- Benefits high quality CPU inference TQ2 on Llama.cpp and Ollama via QAT - Robotcs, Routing, Coding, Multimedia, Advanced tool calling via JiRackDeltaNetTokenizer - JiRack DeltaNet understand video and images that best for Robotics also
Params
27.3 B
Context
262,144
Downloads 30d
494 K
Likes
3
Download history
daily snapshots · 28 days557 K494 K
Sep 27Oct 6Oct 15Oct 24
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 |
|---|---|---|---|---|
| JiRackDeltaNet_27b.Q2_K.gguf | Q2_K | 10.9 GB | 16.5 GB | ✅ Runs comfortably |
| JiRackDeltaNet_27b.Q3_K_M.gguf | Q3_K_M | 13.5 GB | 19.4 GB | ✅ Runs comfortably |
| JiRackDeltaNet_27b.Q4_K_M.gguf | Q4_K_M | 16.8 GB | 23.1 GB | ⚠️ Tight — reduce context |
| JiRackDeltaNet_27b.Q6_K.gguf | Q6_K | 22.4 GB | 29.3 GB | ❌ Won’t fit |
| JiRackDeltaNet_27b.Q8_0.gguf | Q8_0 | 29.0 GB | 36.5 GB | ❌ Won’t fit |
| JiRackDeltaNet_27b.gguf | GGUF | 54.7 GB | 64.7 GB | ❌ Won’t fit |
| model.safetensors | bf16 | 54.6 GB | 64.7 GB | ❌ Won’t fit |
| model.safetensors (bf16, full) | bf16 + 256K ctx | 54.6 GB | 191.7 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$ ollama run jirackdeltanet:27b # pin the quantization explicitly $ ollama run jirackdeltanet:27b-q4_k_m
$ huggingface-cli download CMSManhattan/JiRackDeltaNet_27b-GGUF \
JiRackDeltaNet_27b.Q4_K_M.gguf --local-dir .
$ llama-cli -m JiRackDeltaNet_27b.Q4_K_M.gguf \
-c 8192 -ngl 99 -t 8 --color
$ curl -s https://aimodelscomparison.com/api/v1/models/jirackdeltanet-27b
{
"hf_id": "CMSManhattan/JiRackDeltaNet_27b",
"params_b": 27.32,
"context_length": 262144,
"license": { "id": "mit", "commercial": "yes" },
"downloads_30d": 494347,
"vram_estimates": [
{ "quant": "Q2_K", "gb": 16.5 },
{ "quant": "Q3_K_M", "gb": 19.4 }
],
"updated_at": "2026-10-16T01:00:13Z"
}
Specifications
- Architecture
- Qwen3_5ForCausalLM
- Parameters
- 27.3 B
- Tensor type
- BF16
- Context length
- 262,144
- Vocabulary
- 248,320
- Layers / heads
- 64 / 24
- Licence
- mit
- First seen on the Hub
- 2026-08-29
- Training datasets
- undisclosed
- Added to our catalog
- 2026-09-27
Compare with any text-generation model