Zhipu on NVIDIA Ada

Can I run GLM-4 9B on an RTX 4070?

Yes. GLM-4 9B at Q4_K_M uses 6.5 GB of the 10.5 GB available on a RTX 4070, and runs at about 53.7 tokens per second. You can push the context to 109K.

Runs comfortably

6.5 GB of 10.5 GB · 62%
010 GB
Weights 5.4 GB
KV cache 0.3 GB
Runtime overhead 0.8 GB

GLM-4 9B at Q4_K_M leaves 3.9 GB spare on a RTX 4070. There is room to raise the context length or move up a quantisation level.

Generation53.7tok/s
Prompt processing1296tok/s
Max context109Ktokens
KV per 1K tokens0GB

Every quantisation of GLM-4 9B on a RTX 4070

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

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 17.5 GB 18.6 GB Won't fit — 4.00
INT8 / W8A8 8.50 9.0 GB 10.2 GB Fits, but tight 16K 32.7
Q8_0 (GGUF) 8.50 9.0 GB 10.2 GB Fits, but tight 16K 32.7
FP8 (E4M3) 8.00 8.8 GB 9.9 GB Fits, but tight 23K 33.6
Q6_K 6.56 7.2 GB 8.3 GB Runs comfortably 64K 40.8
AWQ 4-bit 4.25 6.3 GB 7.5 GB Runs comfortably 85K 45.9
GPTQ 4-bit 4.25 6.3 GB 7.5 GB Runs comfortably 85K 45.9
MXFP4 4.25 6.3 GB 7.5 GB Runs comfortably 85K 45.9
Q5_K_M 5.67 6.2 GB 7.3 GB Runs comfortably 88K 47.0
Q5_K_S 5.52 6.0 GB 7.2 GB Runs comfortably 93K 48.2
Q4_K_M 4.85 5.4 GB 6.5 GB Runs comfortably 109K 53.7
Q4_K_S 4.58 5.1 GB 6.3 GB Runs comfortably 116K 56.2
Q4_0 4.55 5.1 GB 6.2 GB Runs comfortably 116K 56.5
IQ4_XS 4.25 4.8 GB 6.0 GB Runs comfortably 124K 59.7
Q3_K_M 3.91 4.5 GB 5.6 GB Runs comfortably 128K 63.8
IQ3_M 3.70 4.3 GB 5.4 GB Runs comfortably 128K 66.6
IQ3_XXS 3.06 3.7 GB 4.8 GB Runs comfortably 128K 76.9
Q2_K 2.63 3.3 GB 4.4 GB Runs comfortably 128K 85.8
IQ2_XXS 2.06 2.8 GB 3.9 GB Runs comfortably 128K 101
IQ1_M 1.75 2.5 GB 3.6 GB Runs comfortably 128K 113

Also worth checking

GLM-4 9B on every GPU
Full hardware tableEverything that runs on a RTX 4070
Full model tableWhy the KV cache is the part that bites
Guide