Models / midashenglm
midashenglm by mispeech
GPU memory needed (smallest version of each): midashenglm 0.6B about 4.8 GB, midashenglm 7B about 9.7 GB. The cheapest runs on RTX 3060 from $0.10/h right now.
9 models (0.6B to 7B) · ↓ 13.9K downloads a month · from $0.10/h on RTX 3060 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
midashenglm 0.6B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-0.6b-fp32 | vLLM | 1.4 GB | 4.8 GB | $0.10/h RTX 3060 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-0804-fp32 | vLLM | 17 GB | 21.1 GB | $0.23/h RTX A5000 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-1021-bf16 | vLLM | 17 GB | 21.1 GB | $0.23/h RTX A5000 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-0804-4bit-bnb | vLLM | 6.2 GB | 9.7 GB | $0.10/h RTX 3060 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-0804-bf16 | vLLM | 17 GB | 21.1 GB | $0.23/h RTX A5000 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-1021-fp32 | vLLM | 17 GB | 21.1 GB | $0.23/h RTX A5000 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-0804-fp8 | vLLM | 10 GB | 13.9 GB | $0.11/h RTX A4000 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-1021-w4a16-gptq | vLLM | 6.9 GB | 10.4 GB | $0.10/h RTX 3060 |
midashenglm 7B
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|
| Originalmispeech/midashenglm-7b-0804-w4a16-gptq | vLLM | 6.9 GB | 10.4 GB | $0.10/h RTX 3060 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").