Serving MiniMax M3 with Triton Inference Server
MiniMax M3 has 428 billion parameters. Triton Inference Server targets industrial deployment across several models. This page gives the memory required and the platforms that qualify.
What is the minimum memory?
MiniMax M3 totals 428 billion parameters. In NVFP4 its weights take about 246 GB including the runtime margin: 428 billion parameters × 0.5 byte (NVFP4) = 214.0 GB of weights; × 1.15 runtime margin = 246.1 GB. In FP8 the footprint doubles, to about 492 GB (428 billion parameters × 1 byte (FP8) = 428.0 GB of weights; × 1.15 runtime margin = 492.2 GB.) The attention cache is not calculated on this page. These models’ architectures (compressed latent, linear or sparse attention, Mamba layers) share no per-token formula; it is measured on the machine, with the target context and concurrency [TO BE MEASURED].
What is Triton Inference Server for?
Triton Inference Server targets industrial deployment across several models. See the Triton Inference Server and MiniMax M3 fact sheets.
On which platforms?
- MiniMax M3 on Mac Studio Ultra
- MiniMax M3 on DGX Station
- MiniMax M3 on RTX PRO 6000 server
- MiniMax M3 on H200 SXM server
- MiniMax M3 on B200 SXM
- MiniMax M3 on B300 SXM
How much memory for MiniMax M3 with Triton Inference Server?
About 246 GB in NVFP4 for the weights, excluding the attention cache.
Does Triton Inference Server suit MiniMax M3?
Triton Inference Server serves MiniMax M3 as soon as the machine offers at least 246 GB of memory, the weight footprint in NVFP4. 7 of the 8 platforms in the catalogue meet that bar.
Method and sources
Memory footprints are calculated, not measured: a reading on real hardware may differ depending on the engine and the exact weight format. Every figure below carries its source and the date of its last check.
- MiniMax M3: 428 billion parameters, 23 billion active, 1,048,576-token context, MiniMax Community licence. Hugging Face repository (config.json, API, model card), checked on 2 September 2026. About 428 billion announced, 23 active; the BF16 repository holds 427 billion elements (854 GB).
- Bytes per parameter: NVFP4 0.5 (4 bits per parameter, hence 0.5 byte; 4-bit published weights weigh 0.54 to 0.56 byte per parameter with their scales (DeepSeek V4 Pro 865 GB for 1,599 billion, Kimi K3 1,561 GB for 2,780 billion)); FP8 1 (8 bits per parameter (E4M3), hence 1 byte, block scales not counted). Source: connaissance/faits.yaml, quantifications family, checked on 2026-08-31.
- Runtime margin × 1.15: QDNA operating assumption, not measured: 15% above the weights for activations, buffers and fragmentation.
- “Fits” threshold at 75% of memory: QDNA assumption, not measured, which keeps the remaining quarter for the attention cache and concurrency.