This model requires you to accept the MIT license terms before download.
Description
Ming-Lite-Omni-1.5 omni understanding (text + image + audio) AND generative image output (--mode image): windowed-ViT vision tower, varlen whisper audio encoder, multi-gate MoE decoder, in-package flow-matching diffusion stack (condition bridge + DiT denoiser + VAE) — 4-mode validated on both paths, seeded byte-reproducible generation
At a glance
| Kind | multimodal model |
|---|---|
| Precision | bfloat16 |
| Container size | 51.1 GiB |
| Weights inside | 50.6 GiB |
| Components | 9 |
| Licence | MIT |
| Origin | inclusionAI/Ming-Lite-Omni-1.5 @ a1085e848350 |
What is inside the container
A .nbx container packs every part of Ming-Lite-Omni-1.5 — 9 components with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| audio | 0 | — | 1.2 GiB |
| linear_proj | 0 | — | 40.0 MiB |
| linear_proj_audio | 0 | — | 23.0 MiB |
| vision | 0 | — | 1.3 GiB |
| model.lm_head | 0 | — | 494 MiB |
| model.model | 0 | 28 | 30.8 GiB |
| condition_connector | 0 | — | 5.8 GiB |
| image_denoiser | 0 | — | 10.7 GiB |
| image_vae | 0 | — | 160 MiB |
Default settings
The values the container ships with. They apply when you do not pass your own.
- Max length
- 32768
- Temperature
- 0
- Top-p
- 1
- Top-k
- 0
Memory needed
The weights of Ming-Lite-Omni-1.5 take 50.6 GiB in bfloat16. That is the floor for the memory the runtime has to place, spread over the devices it finds; activations come on top and grow with the size and length you ask for. A measured peak per device class is not published for this model yet.
Verify what you downloaded
Every download carries its SHA-256, and the file endpoint sends it in the X-NBX-SHA256 header. The container is 51.1 GiB (54,825,244,563 bytes).
$ sha256sum Ming-Lite-Omni-1.5.nbx
70b055166b3e130855952f4da0d670200a43914ed2bb5635e9584d9c8909c752
Questions
What is Ming-Lite-Omni-1.5?
A multimodal model by inclusionAI, distributed as a bfloat16 .nbx container and licensed MIT.
How big is the download?
51.1 GiB, of which 50.6 GiB are weights.
How do I run it?
Install the runtime from the download page, then use the commands under “CLI Usage” below: neurobrix import inclusionAI/Ming-Lite-Omni-1.5, then neurobrix serve --model Ming-Lite-Omni-1.5.
CLI Usage
$ neurobrix import inclusionAI/Ming-Lite-Omni-1.5
$ neurobrix serve --model Ming-Lite-Omni-1.5
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop