Models › GLM-4.7-Flash
GLM-4.7-Flash: VRAM needs and speed by GPU
We have 2 measured speed results for it.
29.94B parameters (mixture-of-experts, ~3.0B active) · 202,752-token context · GGUF published Jan 20, 2026 · GGUF files on Hugging Face
Measured speeds
| Hardware | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Apple M2 Ultra | 59.6 | 1,629 | Q8_0 | Metal FA on | tg32 | llama.cpp benches details |
| NVIDIA DGX Spark (GB10) | 48.7 | 2,364 | Q8_0 | CUDA FA on | tg32 | llama.cpp benches details |
Memory needed per quantization
| Quant | File size | Needs (estimate) | Smallest card it fits |
|---|---|---|---|
| Q4_K | 18.2 GB | 19.8 GB | 24 GB cards |
| Q8_0 | 31.8 GB | 33.4 GB | 48 GB cards |
Expected speed on popular hardware
| Hardware | Best quant that fits | Expected tok/s |
|---|---|---|
| NVIDIA RTX 3060 | Does not fit | |
| NVIDIA RTX 4060 Ti | Does not fit | |
| NVIDIA RTX 3090 | Q4_K | 180 estimate |
| NVIDIA RTX 4090 | Q4_K | 199 estimate |
| NVIDIA RTX 5090 | Q4_K | 315 estimate |
| AMD Radeon RX 7900 XTX | Q4_K | 200 estimate |
| Apple M4 Pro (20-core GPU) (64 GB) | Q8_0 | 36.4 estimate |
| Apple M3 Max (40-core GPU) (64 GB) | Q8_0 | 53.1 estimate |
| AMD Ryzen AI Max+ 395 (Radeon 8060S) | Q8_0 | 32.6 estimate |
Too big for your card?
Run GLM-4.7-Flash on a rented 24-80 GB GPU by the hour.
Rent a bigger GPU on RunPodAffiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.