Workbench / Online
BUILD YOUR COMPUTE STACK
Start with the model. MuseBoard maps the memory and hardware. Settings are mirrored to the URL, so any configuration can be shared.
Compute workbench
B / Estimated VRAM
Inference103.8GB
Serving: weights + KV cache + runtime overhead. Estimate — not a guarantee.
- Model weights93.4 GB
- KV cache1.1 GB
- Runtime overhead9.3 GB
Memory headroom
Free after estimated load
Utilization
Target ≤ 90% of 141.0 GB
Decode ceiling
Theoretical, batch 1, bandwidth-bound
VRAM usage
103.8 GB / 141.0 GB
Recommended configuration
1 × NVIDIA H200 141GB
- Total VRAM
- 141.0 GB
- Topology
- Single GPU
- Interconnect
- NVLink 4 · 900 GB/s
- Board power
- 700 W
Compatibility / 1 × H200
- Ready
Inference
103.8 GB · 74% of 1 × H200
- Ready
Fine-tuning
111.6 GB · 79% of 1 × H200
- Limited
Training
826.7 GB · Needs 8 × H200
Alternative configurations
- Mixture-of-experts: all 46.7B parameters stay resident in memory; only ~12.9B are active per token.
- 01SERVE / READY
Inference
Estimate memory requirements for serving models.
weights + KV cache + overhead
Open in workbench - 02ADAPT / LORA
Fine-tuning
Explore memory requirements for adapting existing models.
frozen base + adapters + activations
Open in workbench - 03TRAIN / ADAM
Training
Understand large-scale compute requirements.
weights + grads + optimizer + activations
Open in workbench
MuseBoard gives first-order estimates for planning conversations. It does not replace profiling, capacity testing or detailed ML infrastructure planning.