← CanIRunLocal · All hardware · All models
Perplexity · 27B parameters · chat, vision.
Ollama weight files: Q4_K_M 17 GB (pplx-decider:27b).
Yes on 64 machines in this list. The smallest is RTX 3060 12GB: Q4_K_M, about 19.3 GB, roughly 5 tokens per second, via Ollama. ollama run pplx-decider:27b
ollama run pplx-decider:27bollama run pplx-decider:27bollama run pplx-decider:27bollama run pplx-decider:27bvllm serve perplexity-ai/pplx-decider-v1.1-27b --max-model-len 8192 --gpu-memory-utilization 0.9| Hardware | Runtime | Quant | Memory | Fit | Speed | Command |
|---|---|---|---|---|---|---|
| RTX 3060 12GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~5 tok/s | ollama run pplx-decider:27b |
| RTX 3080 12GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~5 tok/s | ollama run pplx-decider:27b |
| RTX 4070 12GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~5 tok/s | ollama run pplx-decider:27b |
| RTX 5070 12GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~5 tok/s | ollama run pplx-decider:27b |
| RTX 4060 Ti 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~6 tok/s | ollama run pplx-decider:27b |
| RTX 4070 Ti Super 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~9 tok/s | ollama run pplx-decider:27b |
| RTX 4080 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~9 tok/s | ollama run pplx-decider:27b |
| RTX 5080 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~9 tok/s | ollama run pplx-decider:27b |
| RX 7800 XT 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~8 tok/s | ollama run pplx-decider:27b |
| RX 9070 XT 16GB | Ollama | Q4_K_M | 19.3 GB | partial offload (slower) | ~8 tok/s | ollama run pplx-decider:27b |
| RTX 3090 24GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~33 tok/s | ollama run pplx-decider:27b |
| RTX 4090 24GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~36 tok/s | ollama run pplx-decider:27b |
| RX 7900 XTX 24GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~34 tok/s | ollama run pplx-decider:27b |
| M1 Max (24c GPU) 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M1 Max (32c GPU) 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M1 Pro 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~9 tok/s | ollama run pplx-decider:27b |
| M2 Max (30c GPU) 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M2 Max (38c GPU) 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M2 Pro 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~9 tok/s | ollama run pplx-decider:27b |
| M4 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~6 tok/s | ollama run pplx-decider:27b |
| M3 Max (30c GPU) 36GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~14 tok/s | ollama run pplx-decider:27b |
| M3 Pro 36GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~7 tok/s | ollama run pplx-decider:27b |
| M4 Max (32c GPU) 36GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| Mac Studio M4 Max (32c GPU) 36GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Max (32c GPU) 36GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~22 tok/s | ollama run pplx-decider:27b |
| RTX 5090 32GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~63 tok/s | ollama run pplx-decider:27b |
| M3 Max (40c GPU) 48GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M4 Max (40c GPU) 48GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| M4 Pro 48GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~13 tok/s | ollama run pplx-decider:27b |
| Mac Studio M4 Max (40c GPU) 48GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Max (40c GPU) 48GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~29 tok/s | ollama run pplx-decider:27b |
| M1 Max (32c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M2 Max (30c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M2 Max (38c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M2 Ultra 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~38 tok/s | ollama run pplx-decider:27b |
| M3 Max (40c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M4 Max (40c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| M4 Pro 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~13 tok/s | ollama run pplx-decider:27b |
| Mac Studio M4 Max (40c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Max (40c GPU) 64GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~29 tok/s | ollama run pplx-decider:27b |
| M2 Max (38c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M3 Max (30c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~14 tok/s | ollama run pplx-decider:27b |
| Mac Studio M3 Ultra (60c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~39 tok/s | ollama run pplx-decider:27b |
| Mac Studio M3 Ultra (80c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~39 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Ultra (64c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~56 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Ultra (80c GPU) 96GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~56 tok/s | ollama run pplx-decider:27b |
| M2 Ultra 128GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~38 tok/s | ollama run pplx-decider:27b |
| M3 Max (40c GPU) 128GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~19 tok/s | ollama run pplx-decider:27b |
| M4 Max (40c GPU) 128GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| Mac Studio M4 Max (40c GPU) 128GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~26 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Max (40c GPU) 128GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~29 tok/s | ollama run pplx-decider:27b |
| DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte) | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~12 tok/s | ollama run pplx-decider:27b |
| M2 Ultra 192GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~38 tok/s | ollama run pplx-decider:27b |
| 2x Mac Studio M5 Max (40c GPU) 128GB | MLX | MLX 4bit | 17.6 GB | fits in fast memory | ~45 tok/s | exo # on every Mac (brew install --cask exo); load mlx-community/pplx-decider-v1.1-27b-4bit with Tensor sharding + RDMA |
| Mac Studio M3 Ultra (60c GPU) 256GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~39 tok/s | ollama run pplx-decider:27b |
| Mac Studio M3 Ultra (80c GPU) 256GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~39 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Ultra (64c GPU) 256GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~56 tok/s | ollama run pplx-decider:27b |
| Mac Studio M5 Ultra (80c GPU) 256GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~56 tok/s | ollama run pplx-decider:27b |
| 2x DGX Spark 128GB | vLLM | Official weights | 54.5 GB | fits in fast memory | ~5 tok/s | vllm serve perplexity-ai/pplx-decider-v1.1-27b --tensor-parallel-size 2 --distributed-executor-backend ray --max-model-len 8192 |
| 4x Mac Studio M5 Max (40c GPU) 128GB | MLX | MLX 4bit | 17.6 GB | fits in fast memory | ~70 tok/s | exo # on every Mac (brew install --cask exo); load mlx-community/pplx-decider-v1.1-27b-4bit with Tensor sharding + RDMA |
| 2x Mac Studio M5 Ultra (80c GPU) 256GB | MLX | MLX 4bit | 17.6 GB | fits in fast memory | ~88 tok/s | exo # on every Mac (brew install --cask exo); load mlx-community/pplx-decider-v1.1-27b-4bit with Tensor sharding + RDMA |
| Mac Studio M3 Ultra (80c GPU) 512GB | Ollama | Q4_K_M | 19.3 GB | fits in fast memory | ~39 tok/s | ollama run pplx-decider:27b |
| 4x DGX Spark 128GB | vLLM | Official weights | 54.5 GB | fits in fast memory | ~7 tok/s | vllm serve perplexity-ai/pplx-decider-v1.1-27b --tensor-parallel-size 4 --distributed-executor-backend ray --max-model-len 8192 |
| 4x Mac Studio M5 Ultra (80c GPU) 256GB | MLX | MLX 4bit | 17.6 GB | fits in fast memory | ~138 tok/s | exo # on every Mac (brew install --cask exo); load mlx-community/pplx-decider-v1.1-27b-4bit with Tensor sharding + RDMA |
Estimates, not benchmarks. Model sizes are ollama.com download sizes; memory needed adds headroom for ~8k tokens of context; speeds are rough figures from memory bandwidth. Data updated 2026-10-19.
A project by OnDevice Labs · llms.txt