ai.canshu.com
中文
VRAM Calculator Advisor & TCO API Pricing Compare Models Leaderboard GPUs

Search models, GPUs, and tools

↑↓ Navigate · Enter Open · ★ Watchlist prioritized ai.canshu.com
Q4_K_M: Smooth Fit 19.83 GB / 23.5 GB Usable

Qwen3.8-27B on NVIDIA GeForce RTX 4090 24GB: VRAM Requirement & Speed Estimate

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.8-27B requires 19.83 GB of VRAM (16.85 GB weights + 1.34 GB KV cache + 1.64 GB overhead), fitting within RTX 4090's 23.5 GB usable memory at ~41.4 tokens/s.

Concurrency (Batch):
Dense 27.8B·64L (d=5120)·24Q/4KV·Ctx 256K·Text+Image+Video
Runnable Models →
GPUs
24 GB VRAM·1008 GB/s·Safe Usable 23.5 GB
Smooth Fit19.83 / 23.5 GB~41.4 tok/s
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Smooth Fit
19.83/ 23.5 GB usable

Physical 24 GB · Load 84.3%

Theoretical Decode Speed
~41.4tokens/s

Actual speed varies ~±25% by runtime

Max Context at Batch=1
71K/ 256K

Enable Q8/Q4 KV quantization to scale batch concurrency

Weights: 16.85 GBKV Cache (1×16K): 1.34 GBOverhead: 1.64 GB
🏃 Smooth & Snappy~12.1s for a 500-token response
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run qwen3-8-27b

RTX 4090 Fits Qwen3.8-27B · Cost & TCO

Approx. $1,899 (~$79/GB VRAM) · 450W TDP

📊 All Quantization Levels for Qwen3.8-27B on RTX 4090

Quantization Total VRAM (16K) Verdict Est. Speed Max Context
FP16 / BF16 59.35 GB Out of VRAM (OOM) — —
FP8 / Q8_0 32.77 GB Partial RAM Offload (Slow) ~6.4 tok/s —
GGUF Q6_K 25.89 GB Partial RAM Offload (Slow) ~16.4 tok/s —
GGUF Q5_K_M 22.81 GB Tight Fit (Watch Context) ~32.7 tok/s 26K
Q4_K_M / AWQ 19.83 GB Smooth Fit ~41.4 tok/s 71K
GGUF IQ4_XS 17.7 GB Smooth Fit ~47 tok/s 102K
GGUF Q3_K_M 16.5 GB Smooth Fit ~50.9 tok/s 120K
GGUF Q2_K 13.8 GB Smooth Fit ~62.5 tok/s 160K

💡 Recommended Alternative GPUs & Macs for Qwen3.8-27B

Frequently Asked Questions

Can NVIDIA GeForce RTX 4090 24GB run Qwen3.8-27B locally?

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.8-27B requires 19.83 GB of VRAM (16.85 GB weights + 1.34 GB KV cache + 1.64 GB overhead), fitting within RTX 4090's 23.5 GB usable memory at ~41.4 tokens/s.

What is the best quantization for Qwen3.8-27B on RTX 4090?

The highest-quality quantization that fits cleanly on RTX 4090 is GGUF Q5_K_M (High Balanced Quality), using 22.81 GB VRAM at ~32.7 tokens/s with up to 26K context.

How much VRAM does Qwen3.8-27B need in full FP16 precision?

In unquantized FP16/BF16 precision (16K context), Qwen3.8-27B requires 59.35 GB VRAM (55.6 GB for weights alone).