Apple M2 Ultra meets the minimum VRAM requirement for Q4 inference of codellama/CodeLlama-34b-hf. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
Apple M2 Ultra can run codellama/CodeLlama-34b-hf with Q4 quantization. At approximately 38 tokens/second, you can expect Moderate speed - useful for batch processing.
You have 175GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 17GB | 192GB | 38.18 tok/s | ✅ Fits comfortably |
| Q8 | 35GB | 192GB | 27.68 tok/s | ✅ Fits comfortably |
| FP16 | 70GB | 192GB | 15.23 tok/s | ✅ Fits comfortably |