Can I run OLMoE 1B 7B 0924 locally?
OLMoE 1B 7B 0924 by allenai has 6.9B parameters (1.3B active per token, 64 experts, 8 used per token). At 4-bit with an 8K context it needs about 5.4 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 allenai/OLMoE-1B-7B-0924-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx |
|---|---|---|
| 2-bit (Q2_K) • | rough | 3.4 GB |
| 3-bit (Q3_K_M) • | usable | 4.1 GB |
| 4-bit (Q4_K_M) • | good, most popular | 4.9 GB |
| 5-bit (Q5_K_M) • | very good | 5.6 GB |
| 6-bit (Q6_K) • | near lossless | 6.3 GB |
| 8-bit (Q8_0) • | effectively lossless | 7.9 GB |
| 16-bit (F16/BF16) • | original weights | 13.9 GB |
Which GPUs and Macs run OLMoE 1B 7B 0924?
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 | Q8_0 | 8.4 GB | 264 | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | Q8_0 | 8.4 GB | 211 | details |
| RTX 3090 | 24 GB | Fits | F16 | 14.4 GB | 365 | details |
| RTX 4090 | 24 GB | Fits | F16 | 14.4 GB | 393 | details |
| RTX 5090 | 32 GB | Fits | F16 | 14.4 GB | 699 | details |
| RTX 5070 Ti | 16 GB | Fits | Q8_0 | 8.4 GB | 657 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 14.4 GB | 374 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 14.4 GB | 106 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 14.4 GB | 213 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 14.4 GB | 319 | details |
| DGX Spark | 128 GB | Fits | F16 | 14.4 GB | 106 | details |
| RTX 4070 | 12 GB | Fits | Q8_0 | 8.4 GB | 370 | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | Q8_0 | 8.4 GB | 329 | details |
| RTX 4060 | 8 GB | Fits | Q6_K | 6.8 GB | 258 | details |
| Arc B580 | 12 GB | Fits | Q8_0 | 8.4 GB | 334 | details |
| RTX 3060 Ti | 8 GB | Fits | Q6_K | 6.8 GB | 425 | details |
| RTX 3070 | 8 GB | Fits | Q6_K | 6.8 GB | 425 | details |
| RTX 3080 10GB | 10 GB | Fits | Q8_0 | 8.4 GB | 557 | details |
| RTX 3080 12GB | 12 GB | Fits | Q8_0 | 8.4 GB | 669 | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 14.4 GB | 393 | details |
| RTX 4060 Ti 8GB | 8 GB | Fits | Q6_K | 6.8 GB | 273 | details |
| RTX 4070 Super | 12 GB | Fits | Q8_0 | 8.4 GB | 370 | details |
| RTX 4070 Ti | 12 GB | Fits | Q8_0 | 8.4 GB | 370 | details |
| RTX 4070 Ti Super | 16 GB | Fits | Q8_0 | 8.4 GB | 493 | details |
| RTX 4080 | 16 GB | Fits | Q8_0 | 8.4 GB | 526 | details |
| RTX 4080 Super | 16 GB | Fits | Q8_0 | 8.4 GB | 540 | details |
| RTX 5060 | 8 GB | Fits | Q6_K | 6.8 GB | 425 | details |
| RTX 5070 | 12 GB | Fits | Q8_0 | 8.4 GB | 493 | details |
| RTX 5080 | 16 GB | Fits | Q8_0 | 8.4 GB | 704 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 14.4 GB | 365 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 14.4 GB | 393 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 14.4 GB | 699 | details |
| RTX A6000 | 48 GB | Fits | F16 | 14.4 GB | 299 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 14.4 GB | 374 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 14.4 GB | 699 | details |
| L40S | 48 GB | Fits | F16 | 14.4 GB | 337 | details |
| A100 40GB | 40 GB | Fits | F16 | 14.4 GB | 606 | details |
| A100 80GB | 80 GB | Fits | F16 | 14.4 GB | 795 | details |
| H100 PCIe | 80 GB | Fits | F16 | 14.4 GB | 780 | details |
| H100 SXM | 80 GB | Fits | F16 | 14.4 GB | 1306 | details |
| H200 | 141 GB | Fits | F16 | 14.4 GB | 1871 | details |
| RX 7600 | 8 GB | Fits | Q6_K | 6.8 GB | 273 | details |
| RX 7700 XT | 12 GB | Fits | Q8_0 | 8.4 GB | 317 | details |
| RX 7800 XT | 16 GB | Fits | Q8_0 | 8.4 GB | 458 | details |
| RX 7900 GRE | 16 GB | Fits | Q8_0 | 8.4 GB | 422 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 14.4 GB | 312 | details |
| RX 9070 | 16 GB | Fits | Q8_0 | 8.4 GB | 469 | details |
| RX 9070 XT | 16 GB | Fits | Q8_0 | 8.4 GB | 469 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 14.4 GB | 100 | details |
| Arc A770 16GB | 16 GB | Fits | Q8_0 | 8.4 GB | 411 | details |
| Mac M1 16GB | 16 GB | Fits | Q8_0 | 8.4 GB | 50 | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 14.4 GB | 39 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 14.4 GB | 39 | details |
| Mac M4 16GB | 16 GB | Fits | Q8_0 | 8.4 GB | 88 | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 14.4 GB | 47 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 14.4 GB | 78 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 14.4 GB | 58 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 14.4 GB | 106 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 14.4 GB | 106 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 14.4 GB | 156 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 14.4 GB | 156 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 14.4 GB | 156 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 14.4 GB | 213 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 14.4 GB | 312 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 14.4 GB | 319 | details |
Questions
How much VRAM does OLMoE 1B 7B 0924 need?
About 5.4 GB at 4-bit (Q4_K_M) with an 8K context: 3.9 GB of weights (real GGUF file size), 1.0 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 8.4 GB.
What is the smallest GPU that runs OLMoE 1B 7B 0924?
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 OLMoE 1B 7B 0924 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.
Olmo 3 7B Instruct
7.3B · 7.3 GB at 4-bitOlmo 3 1025 7B
7.3B · 7.3 GB at 4-bitOlmo 3 7B Think
7.3B · 7.3 GB at 4-bitdeepseek coder 7b instruct v1.5
6.9B · 8.3 GB at 4-bitgranite 4.0 h tiny
6.9B · 4.6 GB at 4-bitdeepseek coder 6.7b instruct
6.7B · 8.3 GB at 4-bitModel facts
| Hugging Face repo | allenai/OLMoE-1B-7B-0924 |
| Parameters | 6,919,161,856 (6.9B) · ~1.3B active (computed from config) |
| Layers / hidden size | 16 / 2048 |
| Attention | MHA; 16 heads, 16 KV heads, head dim 128 |
| KV cache per token (FP16) | 128.0 KiB |
| Max context | 4,096 tokens |
| License | apache-2.0 |
| Released on HF | Jul 20, 2024 |
| Downloads (30d) / likes | 45,308 / 159 |
Sources: Hugging Face model API · config.json (allenai/OLMoE-1B-7B-0924) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.