SuperDeepseek V4 Flash Abliterated MQ 2xDGX — Hardware Requirements & GPU Compatibility
ChatReasoning
SuperDeepseek V4 Flash Abliterated MQ 2xDGX is a 306.3B-parameter open language model from Jiunsong in the DeepSeek V4 family. It supports a context window of up to 1,048,576 tokens. At Q4_K_M it needs about 184.07 GB of VRAM — see which GPUs and Macs can run it below.
1.4K downloads
42 likesSpecifications
- Publisher
- Jiunsong
- Family
- DeepSeek V4
- Parameters
- 306.3B
- Architecture
- DeepseekV4ForCausalLM
- Context Length
- 1,048,576 tokens
- Vocabulary Size
- 129,280
- Release Date
- 2026-08-11
- License
- MIT
Get Started
How Much VRAM Does SuperDeepseek V4 Flash Abliterated MQ 2xDGX Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 130.5 GB | 142 GB | 130.16 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 149.6 GB | 161.1 GB | 149.30 GB | 3-bit medium quantization |
| Q4_K_Mest. | 4.80 | 184.1 GB | 195.6 GB | 183.75 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_Mest. | 5.70 | 218.5 GB | 230.1 GB | 218.21 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_Kest. | 6.60 | 253.0 GB | 264.5 GB | 252.66 GB | 6-bit quantization, very good quality |
| Q8_0est. | 8.00 | 306.6 GB | 318.1 GB | 306.25 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 612.8 GB | 624.4 GB | 612.51 GB | Brain floating point 16 — preferred for training |
est.= calculated VRAM estimate; no published GGUF file found for that quantization yet. Other rows are verified against real community uploads.
Which GPUs Can Run SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
Q4_K_M · 184.1 GBSuperDeepseek V4 Flash Abliterated MQ 2xDGX (Q4_K_M) requires 184.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 240+ GB is recommended. Using the full 1049K context window can add up to 11.5 GB, bringing total usage to 195.6 GB. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
Q4_K_M · 184.1 GB6 devices with unified memory can run SuperDeepseek V4 Flash Abliterated MQ 2xDGX, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Studio (M3 Ultra, 256GB).
Runs great
— Plenty of headroom
Decent
— Enough memory, may be tightRelated Models
Frequently Asked Questions
- How much VRAM does SuperDeepseek V4 Flash Abliterated MQ 2xDGX need?
SuperDeepseek V4 Flash Abliterated MQ 2xDGX requires 184.1 GB of VRAM at Q4_K_M, or 612.8 GB at BF16. Full 1049K context adds up to 11.5 GB (195.6 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 306.3B × 4.8 bits ÷ 8 = 183.8 GB
KV Cache + Overhead ≈ 0.3 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 11.8 GB (at full 1049K context)
VRAM usage by quantization
Q4_K_M184.1 GBQ4_K_M + full context195.6 GB- Can NVIDIA GeForce RTX 5090 run SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
No — SuperDeepseek V4 Flash Abliterated MQ 2xDGX requires at least 130.5 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- What's the best quantization for SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
For SuperDeepseek V4 Flash Abliterated MQ 2xDGX, Q4_K_M (184.1 GB) offers the best balance of quality and VRAM usage. Q5_K_M (218.5 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 130.5 GB.
VRAM requirement by quantization
Q2_K130.5 GBQ4_K_M ★184.1 GBQ5_K_M218.5 GBQ6_K253.0 GBQ8_0306.6 GBBF16612.8 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run SuperDeepseek V4 Flash Abliterated MQ 2xDGX on a Mac?
SuperDeepseek V4 Flash Abliterated MQ 2xDGX requires at least 130.5 GB at Q2_K, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run SuperDeepseek V4 Flash Abliterated MQ 2xDGX locally?
Yes — SuperDeepseek V4 Flash Abliterated MQ 2xDGX can run locally on consumer hardware. At Q4_K_M quantization it needs 184.1 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
At Q4_K_M, SuperDeepseek V4 Flash Abliterated MQ 2xDGX can reach ~81 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 184.1 × 0.65 = ~173 tok/s
Estimated speed at Q4_K_M (184.1 GB)
~173 tok/s~173 tok/s~81 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
At Q4_K_M, the download is about 183.75 GB. The full-precision BF16 version is 612.51 GB. The smallest option (Q2_K) is 130.16 GB.
- Which GPUs can run SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
No single consumer GPU has enough VRAM to run SuperDeepseek V4 Flash Abliterated MQ 2xDGX at Q4_K_M (184.1 GB). Multi-GPU or professional hardware is required.
- Which devices can run SuperDeepseek V4 Flash Abliterated MQ 2xDGX?
6 devices with unified memory can run SuperDeepseek V4 Flash Abliterated MQ 2xDGX at Q4_K_M (184.1 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.