# GLM-4.7-Flash (30B-A3B)

> OnPrem World · https://onpremworld.com/en/models/glm-4.7-flash · Last updated: 2026-10-14 · CC BY 4.0

Z.ai's 30B MoE with 3B active and a compressed (MLA) KV cache, under MIT; it fits a 20–24 GB card at 4-bit.

|  |  |
|---|---|
| Developer | Zhipu AI (Z.ai) |
| Parameters | 30B (3B active, MoE) |
| Native context | 202,752 tokens |
| Licence | MIT |
| Commercial use | Yes |
| KV cache | Compressed latent KV (MLA), 47 layers |
| Released | 2026-01 |

```
ollama run glm-4.7-flash:q4_K_M
```

## Memory needed

Weights + KV cache + runtime overhead, in GiB. Lower quantization = less memory, slightly lower quality.

| Quantization | 4K | 8K | 32K |
|---|---|---|---|
| Q4_K_M | 18.5 GiB | 18.7 GiB | 19.9 GiB |
| Q5_K_M | 21.6 GiB | 21.8 GiB | 23.0 GiB |
| Q6_K | 24.9 GiB | 25.1 GiB | 26.3 GiB |
| Q8_0 | 31.9 GiB | 32.1 GiB | 33.3 GiB |
| FP16 | 59.4 GiB | 59.6 GiB | 60.8 GiB |

## Which hardware runs it (8K)

| Hardware | Memory | Best quant | tok/s (est.) |
|---|---|---|---|
| GeForce RTX 5090 | 32 GB | Q6_K | 508 |
| GeForce RTX 4090 | 24 GB | Q5_K_M | 331 |
| Radeon RX 7900 XTX | 24 GB | Q5_K_M | 315 |
| Radeon RX 7900 XT | 20 GB | Q4_K_M | 308 |
| GeForce RTX 3090 | 24 GB | Q5_K_M | 307 |
| RTX 6000 Ada Generation | 48 GB | Q8_0 | 211 |
| RTX PRO 6000 Blackwell Workstation Edition | 96 GB | FP16 | 209 |
| Radeon AI PRO R9700 | 32 GB | Q6_K | 181 |
| Arc Pro B70 | 32 GB | Q6_K | 172 |
| RTX A6000 | 48 GB | Q8_0 | 169 |
| RTX PRO 5000 72GB Blackwell | 72 GB | FP16 | 157 |
| Arc Pro B60 | 24 GB | Q5_K_M | 150 |
| M5 Ultra | 512 GB | FP16 | 130 |
| M5 Max (18-core CPU / 32-core GPU) | 36 GB | Q6_K | 121 |
| M4 Max (14-core CPU / 32-core GPU) | 36 GB | Q6_K | 108 |
| M3 Ultra | 512 GB | FP16 | 89 |
| M5 Max (18-core CPU / 40-core GPU) | 128 GB | FP16 | 67 |
| M5 Pro | 64 GB | Q8_0 | 63 |
| M4 Max (16-core CPU / 40-core GPU) | 128 GB | FP16 | 59 |
| M4 Pro | 64 GB | Q8_0 | 56 |
| M6 (24GB/32GB configurations, 170 GB/s) | 32 GB | Q5_K_M | 52 |
| M5 | 32 GB | Q5_K_M | 47 |
| RTX Spark (N1X) laptops & compact desktops | 128 GB | FP16 | 30 |
| DGX Spark | 128 GB | FP16 | 27 |
| Ryzen AI Max+ PRO 495 (Gorgon Halo) systems | 192 GB | FP16 | 27 |
| Ryzen AI Max+ 395 (Strix Halo) systems | 128 GB | FP16 | 26 |
| GeForce RTX 3060 (12 GB) | 12 GB | Doesn't fit | – |
| GeForce RTX 4060 Ti 16GB | 16 GB | Doesn't fit | – |
| GeForce RTX 4080 SUPER | 16 GB | Doesn't fit | – |
| GeForce RTX 5060 Ti 16GB | 16 GB | Doesn't fit | – |
| GeForce RTX 5070 | 12 GB | Doesn't fit | – |
| GeForce RTX 5070 Ti | 16 GB | Doesn't fit | – |
| GeForce RTX 5080 | 16 GB | Doesn't fit | – |
| Radeon RX 9060 XT 16GB | 16 GB | Doesn't fit | – |
| Radeon RX 9070 XT | 16 GB | Doesn't fit | – |
| Arc B580 | 12 GB | Doesn't fit | – |
| M6 (16GB configuration, 153 GB/s) | 16 GB | Doesn't fit | – |

## Frequently asked questions

### How much VRAM does GLM-4.7-Flash (30B-A3B) need?

With an 8K context, about 18.7 GiB at Q4_K_M and 32.1 GiB at Q8_0 (weights + KV cache + runtime overhead). Longer contexts need more memory for the KV cache.

### Can GLM-4.7-Flash (30B-A3B) run on an GeForce RTX 4090?

Yes. GLM-4.7-Flash (30B-A3B) fits on the GeForce RTX 4090 (24 GB) at Q5_K_M with an 8K context, using about 21.8 GiB, at an estimated 331 tokens/s.

### Can I run GLM-4.7-Flash (30B-A3B) on a Mac?

Yes. GLM-4.7-Flash (30B-A3B) fits on the M3 Ultra (96 GB) at FP16 with an 8K context, using about 59.6 GiB, at an estimated 89 tokens/s.

### Can GLM-4.7-Flash (30B-A3B) be used commercially?

GLM-4.7-Flash (30B-A3B) is released under MIT. The licence allows commercial use (read its conditions).

## Sources

- https://gigazine.net/gsc_news/en/20260120-zai-glm-4-7-flash
- https://huggingface.co/zai-org/GLM-4.7-Flash
- https://huggingface.co/zai-org/GLM-4.7-Flash/resolve/main/config.json
- https://huggingface.co/api/models/zai-org/GLM-4.7-Flash?expand[]=safetensors
- https://ollama.com/library/glm-4.7-flash/tags