Models › Beam 1.0 501B A23B (MoE)
Can I run Beam 1.0 501B A23B (MoE) in Australia?
Not on a single device we track at Q4_K_M. You would need multiple GPUs or a smaller quantisation.
501.27Btotal parameters · 23B active (MoE)
52layers (9 full-attention, 43 sliding-window)
0.39 GBKV cache @ 8k · 8 KV heads × 128 dim
Model card ↗config.json · source of counts
GPU + RAM offload sizing is off for this model: expert formula gave 523.45B > total 501.27B; offload sizing disabled.
Memory needed by quantisation (8k context)
| Quant | Weights | Total needed | Cheapest hardware that fits |
|---|---|---|---|
Q4_K_M | 303.9 GB | 305.3 GB | nothing we track (with a price) |
Q5_K_M | 356.5 GB | 357.9 GB | nothing we track (with a price) |
Q6_K | 411.0 GB | 412.4 GB | nothing we track (with a price) |
Q8_0 | 532.6 GB | 534.0 GB | nothing we track (with a price) |
F16 | 1002.5 GB | 1003.9 GB | nothing we track (with a price) |
All hardware that fits at Q4_K_M
| Hardware | Needs / usable | Est. tok/s | Price | Retailer link |
|---|---|---|---|---|
| Nothing we track fits. | ||||
The tok/s figures are bandwidth-based estimates for this model. Where a device has a measured benchmark, the estimate is calibrated to that run. They are not measurements. Methodology.
Graphics card + system RAM
Offload sizing not available: we couldn't read this model's expert layout from config.json.
Community measured
No reader results yet. Be the first to submit one →
Try other quantisations and context lengths in the Can I run it? tool → · Submit your own Beam 1.0 501B A23B (MoE) benchmark →