← CanIRunLocal · All hardware · All models
HuggingFace · 0.135B parameters · chat.
Ollama weight files: Q5_K_M 0.11 GB (smollm2:135m-instruct-q5_K_M); Q8_0 0.14 GB (smollm2:135m-instruct-q8_0); FP16 0.27 GB (smollm2:135m); Q4_K_M 0.1 GB (smollm2:135m-instruct-q4_K_M).
Yes on 62 machines in this list. The smallest is M1 8GB: FP16, about 0.9 GB, roughly 201 tokens per second, via Ollama. ollama run smollm2:135m
ollama run smollm2:135mollama run smollm2:135mollama run smollm2:135mollama run smollm2:135mlms get unsloth/SmolLM2-135M-Instruct-GGUF@Q8_0llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 -c 8192vllm serve HuggingFaceTB/SmolLM2-135M-Instruct --max-model-len 8192 --gpu-memory-utilization 0.9| Hardware | Runtime | Quant | Memory | Fit | Speed | Command |
|---|---|---|---|---|---|---|
| M1 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~201 tok/s | ollama run smollm2:135m |
| M2 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| M3 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| RTX 3060 Ti 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~996 tok/s | ollama run smollm2:135m |
| RTX 3070 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~996 tok/s | ollama run smollm2:135m |
| RTX 4060 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~604 tok/s | ollama run smollm2:135m |
| RX 7600 8GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~640 tok/s | ollama run smollm2:135m |
| RTX 3080 10GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1689 tok/s | ollama run smollm2:135m |
| M1 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~201 tok/s | ollama run smollm2:135m |
| M2 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| M3 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| M4 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~356 tok/s | ollama run smollm2:135m |
| RTX 3060 12GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~800 tok/s | ollama run smollm2:135m |
| RTX 4070 12GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1120 tok/s | ollama run smollm2:135m |
| RTX 5070 12GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1493 tok/s | ollama run smollm2:135m |
| RTX 4060 Ti 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~640 tok/s | ollama run smollm2:135m |
| RTX 4070 Ti Super 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1493 tok/s | ollama run smollm2:135m |
| RTX 4080 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1593 tok/s | ollama run smollm2:135m |
| RTX 5080 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2133 tok/s | ollama run smollm2:135m |
| RX 7800 XT 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1387 tok/s | ollama run smollm2:135m |
| RX 9070 XT 16GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1422 tok/s | ollama run smollm2:135m |
| M2 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| M3 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~296 tok/s | ollama run smollm2:135m |
| M4 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~356 tok/s | ollama run smollm2:135m |
| RTX 3090 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2080 tok/s | ollama run smollm2:135m |
| RTX 4090 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2240 tok/s | ollama run smollm2:135m |
| RX 7900 XTX 24GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2133 tok/s | ollama run smollm2:135m |
| M1 Pro 32GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~593 tok/s | ollama run smollm2:135m |
| M2 Pro 32GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~593 tok/s | ollama run smollm2:135m |
| M4 32GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~356 tok/s | ollama run smollm2:135m |
| M3 Pro 36GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~444 tok/s | ollama run smollm2:135m |
| Mac Studio M4 Max (32c GPU) 36GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1215 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Max (32c GPU) 36GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1363 tok/s | ollama run smollm2:135m |
| RTX 5090 32GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~3982 tok/s | ollama run smollm2:135m |
| M4 Pro 48GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~809 tok/s | ollama run smollm2:135m |
| Mac Studio M4 Max (40c GPU) 48GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1618 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Max (40c GPU) 48GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1819 tok/s | ollama run smollm2:135m |
| M1 Max 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1185 tok/s | ollama run smollm2:135m |
| M2 Max 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1185 tok/s | ollama run smollm2:135m |
| M3 Max 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1185 tok/s | ollama run smollm2:135m |
| M4 Max 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1618 tok/s | ollama run smollm2:135m |
| Mac Studio M4 Max (40c GPU) 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1618 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Max (40c GPU) 64GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1819 tok/s | ollama run smollm2:135m |
| DGX Spark 64GB (ASUS, Dell, HP, MSI, Acer, Gigabyte) | Ollama | FP16 | 0.9 GB | fits in fast memory | ~758 tok/s | ollama run smollm2:135m |
| Mac Studio M3 Ultra 96GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2427 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Ultra 96GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~3556 tok/s | ollama run smollm2:135m |
| M3 Max 128GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1185 tok/s | ollama run smollm2:135m |
| M4 Max 128GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1618 tok/s | ollama run smollm2:135m |
| Mac Studio M4 Max (40c GPU) 128GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1618 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Max (40c GPU) 128GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~1819 tok/s | ollama run smollm2:135m |
| DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte) | Ollama | FP16 | 0.9 GB | fits in fast memory | ~758 tok/s | ollama run smollm2:135m |
| M2 Ultra 192GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2370 tok/s | ollama run smollm2:135m |
| 2x Mac Studio M5 Max (40c GPU) 128GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~2982 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052 -c 8192 # run rpc-server on each other node |
| Mac Studio M3 Ultra 256GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2427 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Ultra (64c GPU) 256GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~3556 tok/s | ollama run smollm2:135m |
| Mac Studio M5 Ultra (80c GPU) 256GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~3556 tok/s | ollama run smollm2:135m |
| 2x DGX Spark 128GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~1243 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052 -c 8192 # run rpc-server on each other node |
| 4x Mac Studio M5 Max (40c GPU) 128GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~2631 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
| 2x Mac Studio M5 Ultra (80c GPU) 256GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~5829 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052 -c 8192 # run rpc-server on each other node |
| Mac Studio M3 Ultra (80c GPU) 512GB | Ollama | FP16 | 0.9 GB | fits in fast memory | ~2427 tok/s | ollama run smollm2:135m |
| 4x DGX Spark 128GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~1097 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
| 4x Mac Studio M5 Ultra (80c GPU) 256GB | llama.cpp | Q8_0 GGUF | 0.7 GB | fits in fast memory | ~5143 tok/s | llama-server -hf unsloth/SmolLM2-135M-Instruct-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
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 2026-10-18.
A project by OnDevice Labs · llms.txt