Chirp Chirp! 🐦 We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement.
DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
xtools-at·Instruct·released 30 Oct 2026
ollama run hf.co/mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF:Q4_K_MA compact 9B reasoning model distilled from DeepSeek-V4-Pro in Max Effect mode
How much VRAM does DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic need?
At Q4_K_M DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic needs about 7.1 GB with an 8K context (7.1 GB at Q4, 20 GB at full precision). Pick your GPU to see what fits and how fast it runs.
Needs: ≈7.2 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.
Advanced: context 8K · KV cache 16-bit
Smallest GPU for each quant and context (KV cache 16-bit)
| Quantisation | 8K | 16K | 32K | 64K | 128K | 256K |
|---|---|---|---|---|---|---|
| Q3_K_M | ||||||
| Q4_K_M | ||||||
| Q5_K_M | ||||||
| Q6_K | ||||||
| Q8_0 | ||||||
| FP16 |
≤16 GB card24–32 GB48–96 GBmore
| Quantisation | Total |
|---|---|
| Q2_KHeavy lossSmallest, noticeably worse quality | ≈5.3 GB |
| Q3_K_MNoticeable lossSmall, some quality loss | ≈6.1 GB |
| Q4_K_MrecommendedSmall lossRecommended default for local use | ≈7.2 GB |
| Q5_K_MNear-losslessNear-lossless for most tasks | ≈8.0 GB |
| Q6_KNear-losslessPractically lossless | ≈8.9 GB |
| Q8_0LosslessLossless in practice, twice the size of Q4 | ≈11 GB |
| FP16 / BF16LosslessFull precision, as released | ≈20 GB |
Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.
How to pick a quant
Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.
- LosslessF16, Q8_0
- Near-losslessQ6_K, Q5_K_M
- Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
- Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
- Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K
Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B
How much context fits?
At 8K context, Q4_K_M needs ≈7.2 GB(5.4 GB weights + 1.0 GB KV cache + 0.8 GB overhead). Pick a GPU above to see how much context fits.
Weights + overheadKV cache
Can the RTX 4060 run DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic? →
Speed vs quality per quant
Smaller quants run faster and fit in less memory, but lose quality.
Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.
Real-world speed of DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →
No medians yet. Be one of the first 3 to report a setup.
All reports
Run DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic locally
--quantQ4_K_Mdefault; pick your GPU to tailor itmradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF
ollama run hf.co/mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF:Q4_K_M llama-server -hf mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF:Q4_K_MDownloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.
lms get mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF@Q4_K_MThen load it from the LM Studio app or with `lms load`.
vllm serve xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-hereticservices:
vllm:
image: vllm/vllm-openai:latest
command: xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
ports:
- "8000:8000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 8000.
sglang serve xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic --host 0.0.0.0 --port 30000services:
sglang:
image: lmsysorg/sglang:latest
command: sglang serve xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic --host 0.0.0.0 --port 30000
ports:
- "30000:30000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 30000.
from transformers import pipeline
pipe = pipeline("text-generation", model="xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])Quantised versions and downloads
| Repository | Format | Quant | Size |
|---|---|---|---|
| sakamakismile/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-NVFP4 | Safetensors | NVFP4 | — |
| esatapedico/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-NVFP4 | Safetensors | NVFP4 | — |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q2_K | 3.9 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q3_K_S | 4.4 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q3_K_M | 4.7 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q3_K_L | 5.0 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | IQ4_XS | 5.4 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q4_K_S | 5.5 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q4_K_M | 5.8 GB |
| mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF | GGUF | Q5_K_S | 6.5 GB |
About DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
A compact 9B reasoning model distilled from DeepSeek-V4-Pro in Max Effect mode 🔬 DeepSeek-V4-Pro Distillation 🧠 9B Parameters 📐 ~250K Math & STEM Samples ⚡ MTP & NVFP4 Evaluated 💡 What is DeepSeek-V4-Pro-Qwen3.5-9B? DeepSeek-V4-Pro-Qwen3.5-9B is a reasoning-focused fine-tune of Qwen3.5-9B , distilled from responses generated by DeepSeek-V4-Pro in Max Effect mode . The release concentrates its supervised signal on mathematics and STEM problem solving while preserving the efficiency and practical deployability of the 9B parameter class. 🧩 Structured Reasoning Learns rigorous problem decomposition and multi-step solution patterns from a strong teacher. 📐 Math & STEM Focus Trained on approximately 250,000 samples centered on mathematics and scientific reasoning. 🧪 Cross-Domain Transfer Shows small post-training gains in programming and tool use despite receiving no coding-specific SFT data. ⚙️ Practical Deployment Evaluated through vLLM NVFP4 and llama.cpp MTP-Q8_0 deployment paths. 🧪 Hardware Cooperation & Joint Collaboration This project is built in close collaboration with hardware engineer Kyle Hessling , whose infrastructure, training support, and evaluation assistance helped make this release possible. 👉 Follow his hardware and model-training updates on X / Twitter: @KyleHessling1 🦥 Fine-tuning Framework (Unsloth)
Full model card on Hugging Face →Alternatives to DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
All alternatives →Aloha! 🌺 Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding.
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems.
K2-Horizon-7B is the medium dense member of the K2-Horizon family: a 7B-core decoder-only model with a 512K context window.
Model Summary: Granite-4.1-8B is a 8B parameter long-context instruct model finetuned from Granite-4.1-8B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets.
DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic FAQ
- How much VRAM does DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic need?
- About 7.1 GB at Q4_K_M with an 8K context. Q8_0 needs about 11 GB and full precision about 20 GB. Longer contexts need more memory for the KV cache.
- Can DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic run on the RTX 4060?
- Yes. It fits at Q4_K_M (about 7.2 GB), at an estimated 26 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.
- Can I use DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic commercially?
- Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.
- What is the context length of DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic?
- 262,144 tokens (256K). The full window needs far more memory than an 8K context.
- How do I run DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic with Ollama or llama.cpp?
- With Ollama: ollama run hf.co/mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF:Q4_K_M. With llama.cpp: llama-server -hf mradermacher/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic-GGUF:Q4_K_M, which starts an OpenAI-compatible server.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments