GLM-5.2 on your hardware. No egress, no cloud, no external provider.
Request deploymentDeployment includes 4x NVIDIA GB10 variants running GLM-5.2, accessible via a local inference endpoint. No requests leave the customer network.
| Model | GLM-5.2 |
| Hardware | 4x NVIDIA GB10 |
| Throughput | ~30 tokens/s |
| Egress | None |
One-time deployment fee.
Request a deployment or ask a question.