Can OLMoE 1B 7B 0924 run on the DGX Spark?
Yes. At 8K context the best fit is F16, using about 14.4 GB of the 109 GB usable on the DGX Spark. Estimated speed ceiling: ~106 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | Max tok/s (est., 8K) |
|---|---|---|
| 2-bit (Q2_K) • | 3.4 GB Fits | 575 |
| 3-bit (Q3_K_M) • | 4.1 GB Fits | 441 |
| 4-bit (Q4_K_M) • | 4.9 GB Fits | 350 |
| 5-bit (Q5_K_M) • | 5.6 GB Fits | 299 |
| 6-bit (Q6_K) • | 6.3 GB Fits | 259 |
| 8-bit (Q8_0) • | 7.9 GB Fits | 200 |
| 16-bit (F16/BF16) • | 13.9 GB Fits | 106 |
• real GGUF file size from allenai/OLMoE-1B-7B-0924-GGUF. Speed is a bandwidth ceiling (273 GB/s), not a benchmark. Methodology.
What fits instead
- Cheapest listed GPU that runs OLMoE 1B 7B 0924: Arc B580 (12 GB, launch MSRP $249) at Q8_0, fully in VRAM at 8K context.
- Largest popular model that fits the DGX Spark: gpt-oss 120B (117B, MoE) at F16, about 61.9 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
OLMoE 1B 7B 0924 on every GPU and Mac · Every model that runs on the DGX Spark · Other models that fit 64 GB VRAM · Try other settings in the calculator
Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs
Data: Hugging Face Hub (allenai/OLMoE-1B-7B-0924), NVIDIA specs. Model data last checked 2026-11-18; refreshed hourly.