Qwen on NVIDIA Ada

Can I run Qwen3 8B on an RTX 4070?

Yes. Qwen3 8B at Q4_K_M uses 6.7 GB of the 10.5 GB available on a RTX 4070, and runs at about 56.5 tokens per second. You can push the context to 35K.

Runs comfortably

6.7 GB of 10.5 GB · 64%
010 GB
Weights 4.7 GB
KV cache 1.1 GB
Runtime overhead 0.8 GB

Qwen3 8B at Q4_K_M leaves 3.8 GB spare on a RTX 4070. There is room to raise the context length or move up a quantisation level.

Generation56.5tok/s
Prompt processing1487tok/s
Max context35Ktokens
KV per 1K tokens0GB

Every quantisation of Qwen3 8B on a RTX 4070

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 15.3 GB 17.2 GB Won't fit — 4.70
INT8 / W8A8 8.50 7.8 GB 9.8 GB Fits, but tight 13K 35.8
Q8_0 (GGUF) 8.50 7.8 GB 9.8 GB Fits, but tight 13K 35.8
FP8 (E4M3) 8.00 7.6 GB 9.6 GB Fits, but tight 14K 36.6
Q6_K 6.56 6.3 GB 8.2 GB Runs comfortably 24K 43.9
AWQ 4-bit 4.25 5.8 GB 7.7 GB Runs comfortably 28K 47.4
GPTQ 4-bit 4.25 5.8 GB 7.7 GB Runs comfortably 28K 47.4
MXFP4 4.25 5.8 GB 7.7 GB Runs comfortably 28K 47.4
Q5_K_M 5.67 5.4 GB 7.4 GB Runs comfortably 30K 50.1
Q5_K_S 5.52 5.3 GB 7.2 GB Runs comfortably 31K 51.3
Q4_K_M 4.85 4.7 GB 6.7 GB Runs comfortably 35K 56.5
Q4_K_S 4.58 4.5 GB 6.4 GB Runs comfortably 37K 58.9
Q4_0 4.55 4.5 GB 6.4 GB Runs comfortably 37K 59.2
IQ4_XS 4.25 4.2 GB 6.2 GB Runs comfortably 39K 62.2
Q3_K_M 3.91 4.0 GB 5.9 GB Runs comfortably 41K 65.9
IQ3_M 3.70 3.8 GB 5.7 GB Runs comfortably 42K 68.4
IQ3_XXS 3.06 3.3 GB 5.2 GB Runs comfortably 45K 77.5
Q2_K 2.63 2.9 GB 4.9 GB Runs comfortably 48K 85.1
IQ2_XXS 2.06 2.5 GB 4.4 GB Runs comfortably 51K 97.8
IQ1_M 1.75 2.2 GB 4.2 GB Runs comfortably 53K 106

Also worth checking

Qwen3 8B on every GPU
Full hardware tableEverything that runs on a RTX 4070
Full model tableWhy the KV cache is the part that bites
Guide