MiMo-V2.5-Pro
Official · FP8 · safetensors
Runtime certified: SGLang vLLM
| Precision / quantization | Format | Bits | Publisher trust | Deployment state |
|---|---|---|---|---|
| FP8 | safetensors | 8 | Official | Runtime certified |
MiMo · General · MiMo V2.5
MiMo-V2.5-Pro: Specs, GPU requirements, hosting & deployment
MiMo-V2.5-Pro is a current General release in the MiMo family, certified from its immutable official source revision.
Total parameters1,020.0B
Active parameters42.0B
Context window1,048,576 tokens
Release statusCurrent · Current
ArchitectureMiMoV2ForCausalLM
ModalitiesText
LicenseMIT License
Deployment stateDeployable with certified configuration
Commercial stateCommercial Use Allowed
Release date2026-04-27
Architecture, strengths and use cases
MiMo-V2.5-Pro is an official self-hostable Xiaomi release in the MiMo General track. It uses MiMoV2ForCausalLM, carries 1020B / 42B MoE, supports a verified context window of 1,048,576 tokens, and exposes only source-certified checkpoints and runtime combinations. Its release, license and deployment facts are tied to the official repository revision rather than inferred from the family name.
MiMo-V2.5-Pro uses MiMoV2ForCausalLM with 1020B / 42B MoE. The stored architecture profile records 70 layers, hidden width 6144, 128 attention heads and a 1,048,576 token context ceiling where the official configuration publishes those fields.
Verified MiMoV2ForCausalLM architecture
1,048,576 token certified context
Official FP8 checkpoint coverage
Private LLM inference
Agentic workflows
Long-context document processing
Certified checkpoints and quantizations
Only exact, source-certified variants attached to this release are shown.
Recommended hardware
Use only the listed sglang, vllm runtime configuration with the exact checkpoint precision. The recommendation engine checks weight memory, KV cache, runtime overhead, safety headroom, GPU architecture and topology before ranking rentable and reference hardware.
Estimated weight memory978.45 GB
KV cache1.64 GB
Runtime overhead233.06 GB
Total with safety margin1504.3 GB
Technically compatible reference sizing
No current FusionPointAI inventory passes every runtime, architecture, topology and memory gate for this exact checkpoint. Use the estimator for global compatible hardware.
License and self-hosting considerations
The full 1020B resident weight set must fit across the selected GPUs. Context and concurrency add KV-cache memory; MoE active parameters do not replace resident-weight memory.
MiMo-V2.5-Pro is published under MIT License. The release-scoped official license link and verification date are retained with the catalogue record.
Commercial Use Allowed. The official license permits commercial use; its obligations still apply.
Alternatives and sibling variants
MiMo-V2.5-Pro
Release timeline
xiaomi-mimo/mimo/mimo-v2-5-pro
Source details
Source revision: c5e3f3d5a12c6bf0d7db7afd82ae50b7889f7aac57816568dbed4967d2c1c0ad
XiaomiMiMo/MiMo-V2.5-Pro @ 2026-07-18T21:51:21Z
Frequently asked questions
How much VRAM does MiMo-V2.5-Pro require?
The exact estimate depends on checkpoint precision, context and concurrency; the page recommendation uses the certified memory engine and compatible runtime configurations.
Which runtimes support MiMo-V2.5-Pro?
Certified upstream runtime evidence is available for sglang, vllm.
Can MiMo-V2.5-Pro be used commercially?
The stored commercial classification is Commercial Use Allowed. Consult the official MIT License terms for the final obligations.
Official references
Last verified: 2026-08-26