← CanIRunLocal · All hardware · All models
Meta · 6.74B parameters · coding.
Ollama weight files: Q5_K_M 4.8 GB (codellama:7b-instruct-q5_K_M); Q8_0 7.2 GB (codellama:7b-instruct-q8_0); Q4_0 3.8 GB (codellama:7b); Q4_K_M 4.1 GB (codellama:7b-instruct-q4_K_M).
Yes on 58 machines in this list. The smallest is M1 8GB: Q4_K_M, about 5 GB, roughly 13 tokens per second, via Ollama. ollama run codellama:7b-instruct-q4_K_M
ollama run codellama:7b-instruct-q8_0ollama run codellama:7b-instruct-q8_0ollama run codellama:7b-instruct-q8_0ollama run codellama:7b-instruct-q8_0vllm serve meta-llama/CodeLlama-7b-Instruct-hf --max-model-len 8192 --gpu-memory-utilization 0.9| Hardware | Runtime | Quant | Memory | Fit | Speed | Command |
|---|---|---|---|---|---|---|
| M1 8GB | Ollama | Q4_K_M | 5 GB | fits in fast memory | ~13 tok/s | ollama run codellama:7b-instruct-q4_K_M |
| M2 8GB | Ollama | Q4_K_M | 5 GB | fits in fast memory | ~20 tok/s | ollama run codellama:7b-instruct-q4_K_M |
| M3 8GB | Ollama | Q4_K_M | 5 GB | fits in fast memory | ~20 tok/s | ollama run codellama:7b-instruct-q4_K_M |
| RTX 3060 Ti 8GB | Ollama | Q5_K_M | 5.7 GB | fits in fast memory | ~56 tok/s | ollama run codellama:7b-instruct-q5_K_M |
| RTX 3070 8GB | Ollama | Q5_K_M | 5.7 GB | fits in fast memory | ~56 tok/s | ollama run codellama:7b-instruct-q5_K_M |
| RTX 4060 8GB | Ollama | Q5_K_M | 5.7 GB | fits in fast memory | ~34 tok/s | ollama run codellama:7b-instruct-q5_K_M |
| RX 7600 8GB | Ollama | Q5_K_M | 5.7 GB | fits in fast memory | ~36 tok/s | ollama run codellama:7b-instruct-q5_K_M |
| RTX 3080 10GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~63 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M1 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~8 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M2 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~11 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M3 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~11 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~13 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 3060 12GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~30 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 4070 12GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~42 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 5070 12GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~56 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 4060 Ti 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~24 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 4070 Ti Super 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~56 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 4080 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~60 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 5080 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~80 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RX 7800 XT 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~52 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RX 9070 XT 16GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~53 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M2 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~11 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M3 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~11 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~13 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 3090 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~78 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 4090 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~84 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RX 7900 XTX 24GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~80 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M1 Pro 32GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~22 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M2 Pro 32GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~22 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 32GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~13 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M3 Pro 36GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~17 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M4 Max (32c GPU) 36GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~46 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Max (32c GPU) 36GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~51 tok/s | ollama run codellama:7b-instruct-q8_0 |
| RTX 5090 32GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~149 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 Pro 48GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~30 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M4 Max (40c GPU) 48GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~61 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Max (40c GPU) 48GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~68 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M1 Max 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~44 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M2 Max 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~44 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M3 Max 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~44 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 Max 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~61 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M4 Max (40c GPU) 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~61 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Max (40c GPU) 64GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~68 tok/s | ollama run codellama:7b-instruct-q8_0 |
| DGX Spark 64GB (ASUS, Dell, HP, MSI, Acer, Gigabyte) | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~28 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M3 Ultra 96GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~91 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Ultra 96GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~133 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M3 Max 128GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~44 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M4 Max 128GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~61 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M4 Max (40c GPU) 128GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~61 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Max (40c GPU) 128GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~68 tok/s | ollama run codellama:7b-instruct-q8_0 |
| DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte) | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~28 tok/s | ollama run codellama:7b-instruct-q8_0 |
| M2 Ultra 192GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~89 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M3 Ultra 256GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~91 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Ultra (64c GPU) 256GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~133 tok/s | ollama run codellama:7b-instruct-q8_0 |
| Mac Studio M5 Ultra (80c GPU) 256GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~133 tok/s | ollama run codellama:7b-instruct-q8_0 |
| 2x DGX Spark 128GB | vLLM | Official weights | 14.4 GB | fits in fast memory | ~19 tok/s | vllm serve meta-llama/CodeLlama-7b-Instruct-hf --tensor-parallel-size 2 --distributed-executor-backend ray --max-model-len 8192 |
| Mac Studio M3 Ultra (80c GPU) 512GB | Ollama | Q8_0 | 8.1 GB | fits in fast memory | ~91 tok/s | ollama run codellama:7b-instruct-q8_0 |
| 4x DGX Spark 128GB | vLLM | Official weights | 14.4 GB | fits in fast memory | ~27 tok/s | vllm serve meta-llama/CodeLlama-7b-Instruct-hf --tensor-parallel-size 4 --distributed-executor-backend ray --max-model-len 8192 |
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 2026-10-16.
A project by OnDevice Labs · llms.txt