estimate, except measured file sizes
Can I run Reflection Beam 1.0 locally?
501.3B total, 23.4B active. Parameter count is the published figure.
Text-only MoE, 256 experts, 8 active. Ollama offers beam:cloud only (no local tag on 17 Oct 2026). Official serving reference: 4× NVIDIA B300. Speed is not estimated: the formula streams every weight and understates sparse MoE.
Parameter source · Architecture source · Model page
speed: not estimated for MoE yet
Memory per quant
KV is at 8,192 tokens. Overhead is 0.8 GB. Both are estimates. A measured file size is labelled. The last column is the cheapest machine that fits the total fully.
| Quant | File size | KV at 8K | +0.8 overhead | Total | Cheapest full fit |
|---|---|---|---|---|---|
| Q8_0 | 532.8 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 534.0 GB est. | Nothing on our list of 29 machines fits this quant. |
| UD-Q6_K_XL | 453.2 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 454.4 GB est. | Nothing on our list of 29 machines fits this quant. |
| UD-Q5_K_XL | 369.5 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 370.7 GB est. | Nothing on our list of 29 machines fits this quant. |
| UD-Q4_K_XL | 306.5 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 307.7 GB est. | Nothing on our list of 29 machines fits this quant. |
| UD-Q3_K_XL | 242.3 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 243.5 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-IQ3_XXS | 197.2 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 198.3 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-Q2_K_XL | 186.9 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 188.1 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-IQ2_M | 164.8 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 166.0 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-IQ2_XXS | 133.9 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 135.1 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-IQ1_M | 123.6 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 124.8 GB est. | Mac Studio M5 Ultra 256GB, €12,500 (estimate) |
| UD-IQ1_S | 103.1 GB measured GGUF | 0.4 GB est. | 0.8 GB est. | 104.3 GB est. | Mac Studio M4 Max 128GB, €3,290 (estimate) |
Cheapest rig that fits
Q4, full: Nothing on our list of 29 machines fits this quant.
Q4, slow, partly in system RAM: Nothing on our list of 29 machines fits this quant.
UD-IQ1_S (smallest file), full: Mac Studio M4 Max 128GB, €3,290 (estimate)
UD-IQ1_S (smallest file), slow, partly in system RAM: Dual RTX 3090, 48 GB, slow, partly in system RAM, €5,408 (estimate)
On a specific GPU
- Reflection Beam 1.0 on RTX 3060 12GB
- Reflection Beam 1.0 on RTX 4060 Ti 16GB
- Reflection Beam 1.0 on RTX 3090
- Reflection Beam 1.0 on RTX 4090
- Reflection Beam 1.0 on RTX 5080
- Reflection Beam 1.0 on RTX 5090
- Reflection Beam 1.0 on RX 7900 XTX
- Reflection Beam 1.0 on Dual RTX 3090
- Reflection Beam 1.0 on Dual RTX 4090