MiMo V2 Flash
Sparse compute, substantial memory. A closer look at the economics of a large MoE.
- Published size
- 309B
- Active parameters
- 15B
- Context window
- 256K
- Architecture
- MoE
- License
- MIT
Deployment considerations
15B active parameters does not mean 15B worth of weights. Size a fully resident deployment for the total parameter count.
The model card reports 309B total parameters and 15B active parameters per token.
The published checkpoint uses FP8. Our bit-width calculator is a theoretical comparison, not a list of validated checkpoint formats.
Hybrid attention and multi-token prediction make runtime support especially important. Use the upstream deployment recipe and validate your exact configuration.
Give your model
some breathing room.
Weights meet or exceed this budget. Consider more memory or a supported lower-precision checkpoint.
Decimal GB; nominal parameter counts where marked ~. Bit widths illustrate weight storage, not validated quantizations. Understand the estimate
No invented leaderboards.
Latency, throughput, and cost per token will appear here after a reproducible run. Until then, this page helps you understand the model—not predict its performance.
MiMo V2 Flash deployment FAQ
How much GPU memory does MiMo V2 Flash need?
At 16-bit precision, the estimated weight storage is 618 GB. At 8-bit it is 309 GB, and at 4-bit it is 154.5 GB. These are theoretical weight-only estimates, excluding KV cache, runtime allocations, and quantization metadata. A working deployment needs additional memory and a supported checkpoint.
Has BenchGrid benchmarked MiMo V2 Flash?
Not yet. This profile contains publisher specifications and calculated weight-memory estimates. We do not currently publish measured latency, throughput, or cost per token for this model.
Where do these specifications come from?
The specifications are based on the official Xiaomi model card linked on this page. Memory estimates use the stated total parameter count, including inactive experts for MoE models. Nominal model sizes are labeled with ~.
Explore your compute options.
Check available hardware, quotas, and current pricing with the provider. These links are not verified deployments or performance recommendations.
