Can I run SmolLM2 360M locally?
SmolLM2 360M by HuggingFaceTB has 362M parameters. At 4-bit with an 8K context it needs about 1.1 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 8 GB of VRAM.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from mradermacher/SmolLM2-360M-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx |
|---|---|---|---|
| 2-bit (Q2_K) • | rough | 0.9 GB | 1.0 GB |
| 3-bit (Q3_K_M) • | usable | 0.9 GB | 1.0 GB |
| 4-bit (Q4_K_M) • | good, most popular | 0.9 GB | 1.1 GB |
| 5-bit (Q5_K_M) • | very good | 0.9 GB | 1.1 GB |
| 6-bit (Q6_K) • | near lossless | 1.0 GB | 1.1 GB |
| 8-bit (Q8_0) • | effectively lossless | 1.0 GB | 1.2 GB |
| 16-bit (F16/BF16) • | original weights | 1.3 GB | 1.5 GB |
Which GPUs and Macs run SmolLM2 360M?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
| Hardware | Memory | Fits? | Best quant | Needs | Max tok/s (est.) | |
|---|---|---|---|---|---|---|
| RTX 3060 12GB | 12 GB | Fits | F16 | 1.5 GB | 496 | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | F16 | 1.5 GB | 397 | details |
| RTX 3090 | 24 GB | Fits | F16 | 1.5 GB | 1290 | details |
| RTX 4090 | 24 GB | Fits | F16 | 1.5 GB | 1389 | details |
| RTX 5090 | 32 GB | Fits | F16 | 1.5 GB | 2470 | details |
| RTX 5070 Ti | 16 GB | Fits | F16 | 1.5 GB | 1235 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 1.5 GB | 1323 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 1.5 GB | 376 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 1.5 GB | 753 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 1.5 GB | 1129 | details |
| DGX Spark | 128 GB | Fits | F16 | 1.5 GB | 376 | details |
| RTX 4070 | 12 GB | Fits | F16 | 1.5 GB | 695 | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | F16 | 1.5 GB | 617 | details |
| RTX 4060 | 8 GB | Fits | F16 | 1.5 GB | 375 | details |
| Arc B580 | 12 GB | Fits | F16 | 1.5 GB | 628 | details |
| RTX 3060 Ti | 8 GB | Fits | F16 | 1.5 GB | 617 | details |
| RTX 3070 | 8 GB | Fits | F16 | 1.5 GB | 617 | details |
| RTX 3080 10GB | 10 GB | Fits | F16 | 1.5 GB | 1047 | details |
| RTX 3080 12GB | 12 GB | Fits | F16 | 1.5 GB | 1257 | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 1.5 GB | 1389 | details |
| RTX 4060 Ti 8GB | 8 GB | Fits | F16 | 1.5 GB | 397 | details |
| RTX 4070 Super | 12 GB | Fits | F16 | 1.5 GB | 695 | details |
| RTX 4070 Ti | 12 GB | Fits | F16 | 1.5 GB | 695 | details |
| RTX 4070 Ti Super | 16 GB | Fits | F16 | 1.5 GB | 926 | details |
| RTX 4080 | 16 GB | Fits | F16 | 1.5 GB | 988 | details |
| RTX 4080 Super | 16 GB | Fits | F16 | 1.5 GB | 1014 | details |
| RTX 5060 | 8 GB | Fits | F16 | 1.5 GB | 617 | details |
| RTX 5070 | 12 GB | Fits | F16 | 1.5 GB | 926 | details |
| RTX 5080 | 16 GB | Fits | F16 | 1.5 GB | 1323 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 1.5 GB | 1290 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 1.5 GB | 1389 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 1.5 GB | 2470 | details |
| RTX A6000 | 48 GB | Fits | F16 | 1.5 GB | 1059 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 1.5 GB | 1323 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 1.5 GB | 2470 | details |
| L40S | 48 GB | Fits | F16 | 1.5 GB | 1191 | details |
| A100 40GB | 40 GB | Fits | F16 | 1.5 GB | 2143 | details |
| A100 80GB | 80 GB | Fits | F16 | 1.5 GB | 2810 | details |
| H100 PCIe | 80 GB | Fits | F16 | 1.5 GB | 2757 | details |
| H100 SXM | 80 GB | Fits | F16 | 1.5 GB | 4617 | details |
| H200 | 141 GB | Fits | F16 | 1.5 GB | 6616 | details |
| RX 7600 | 8 GB | Fits | F16 | 1.5 GB | 397 | details |
| RX 7700 XT | 12 GB | Fits | F16 | 1.5 GB | 595 | details |
| RX 7800 XT | 16 GB | Fits | F16 | 1.5 GB | 860 | details |
| RX 7900 GRE | 16 GB | Fits | F16 | 1.5 GB | 794 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 1.5 GB | 1103 | details |
| RX 9070 | 16 GB | Fits | F16 | 1.5 GB | 882 | details |
| RX 9070 XT | 16 GB | Fits | F16 | 1.5 GB | 882 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 1.5 GB | 353 | details |
| Arc A770 16GB | 16 GB | Fits | F16 | 1.5 GB | 772 | details |
| Mac M1 16GB | 16 GB | Fits | F16 | 1.5 GB | 94 | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 1.5 GB | 138 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 1.5 GB | 138 | details |
| Mac M4 16GB | 16 GB | Fits | F16 | 1.5 GB | 165 | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 1.5 GB | 165 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 1.5 GB | 276 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 1.5 GB | 207 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 1.5 GB | 376 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 1.5 GB | 376 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 1.5 GB | 551 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 1.5 GB | 551 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 1.5 GB | 551 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 1.5 GB | 753 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 1.5 GB | 1103 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 1.5 GB | 1129 | details |
Questions
How much VRAM does SmolLM2 360M need?
About 1.1 GB at 4-bit (Q4_K_M) with an 8K context: 0.3 GB of weights (real GGUF file size), 0.3 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 1.2 GB.
What is the smallest GPU that runs SmolLM2 360M?
At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 (8 GB).
Can I run SmolLM2 360M on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M1 16GB or larger, using the default macOS GPU memory limit.
Similar models
Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.
SmolLM2 360M Instruct
362M · 1.1 GB at 4-bitSmolVLM2 500M Video Instruct
507M · 1.1 GB at 4-bitSmolVLM 500M Instruct
507M · 1.1 GB at 4-bitLFM2.5 350M
354M · 0.8 GB at 4-bitLFM2 350M
354M · 1.0 GB at 4-bitLFM2.5 VL 450M
449M · 0.8 GB at 4-bitModel facts
| Hugging Face repo | HuggingFaceTB/SmolLM2-360M |
| Parameters | 361,821,120 (362M) |
| Layers / hidden size | 32 / 960 |
| Attention | GQA; 15 heads, 5 KV heads, head dim 64 |
| KV cache per token (FP16) | 40.0 KiB |
| Max context | 8,192 tokens |
| License | apache-2.0 |
| Released on HF | Oct 31, 2024 |
| Downloads (30d) / likes | 135,933 / 147 |
Sources: Hugging Face model API · config.json (HuggingFaceTB/SmolLM2-360M) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.