README.md

---
license: apache-2.0
license_link: https://www.apache.org/licenses/LICENSE-2.0
pipeline_tag: image-text-to-text
tags:
- conversational
- image-text-to-text
- audio-text-to-text
- moe
base_model: thinkingmachines/Inkling
---

# Inkling - Independent Apache 2.0 Mirror

This repository is an independently operated mirror of the Inkling model
materials originally published by Thinking Machines Lab.

## License

The model materials in this repository are distributed under the Apache License,
Version 2.0. See [`LICENSE`](./LICENSE).

This repository operator imposes no additional acceptable-use, field-of-use, or
redistribution restrictions on the rights granted under the Apache License 2.0.

## Provenance

Upstream repository:

`https://huggingface.co/thinkingmachines/Inkling`

Upstream revision:

`e4aa5ee880fbb0d2c1a93b1e4a39f2d4b97eb28a`

This mirror was created from Hugging Face's server-side repository duplication
flow. For model repositories, Hugging Face documents that this operation
squashes the source Git history into a single commit.

The upstream publisher separately publishes a Model Acceptable Use Policy. That
policy is not incorporated into the Apache License contained in this repository
and is not presented by this mirror as a condition of access, download, use,
modification, or redistribution.

This statement does not address any separate agreement an individual recipient
may have entered into directly with the upstream publisher.

## Modifications

This README and the repository metadata were created or modified by the operator
of this mirror. A complete Apache 2.0 `LICENSE` file was added.

Unless otherwise identified, the model artifact files correspond to the upstream
revision listed above.

At the upstream revision listed above, the source repository did not contain a
separate `NOTICE` file or standalone `LICENSE` file. Any copyright, attribution,
or notice material embedded in the model artifact, tokenizer, template, or
configuration files is preserved in those files.

## Model Summary

Inkling is a general-purpose multimodal autoregressive transformer that accepts
text, image, and audio inputs and generates text outputs. The upstream model card
describes it as a 975B-parameter sparse Mixture-of-Experts model with 41B active
parameters.

For implementation details, supported local serving stacks, benchmark claims,
safety notes, and limitations, consult the upstream model card at the pinned
revision above.

## Independence And Trademarks

This repository is not affiliated with, sponsored by, or endorsed by Thinking
Machines Lab.

"Thinking Machines Lab" and "Inkling" are used only to identify the source and
subject of the mirrored materials. No trademark rights are claimed.