KARAKURI Open Models: Model List and Licenses
About KARAKURI
KARAKURI is a Japanese company. On its karakuri-ai organization on Hugging Face it releases language and vision-language models for Japanese and English.
It builds on strong overseas models and further trains them for Japanese.
What It Releases
- KARAKURI LM (2024): language models including a 70B based on Llama 2 70B, an 8x7B based on Swallow-MX-8x7b from the Institute of Science Tokyo, and 7B models based on Mistral-7B.
- karakuri-lm-32b-thinking-2501-exp (January 2025): an experimental 32B based on QwQ-32B-Preview with stronger reasoning.
- KARAKURI VL (July 2025): 32B models based on Qwen2.5-VL-32B-Instruct that also read images (instruct and an experimental thinking version).
- KARAKURI VL 2 8B Thinking 2603 (March 2026): according to its model card, a Japanese and English vision-language model fine-tuned from Qwen3-VL-8B-Thinking.
Licenses
The models from 2025 onward (VL 2, VL, the 32B thinking LM) and the 8x7B and 7B-apm-v0.2 are Apache-2.0. The 70B KARAKURI LM uses a custom license (other), and 7B-apm-v0.1 uses the Gemma license. Read each model’s license text before use. This summary is not legal advice.
Running Them Locally
- There are no official GGUF builds (as of 2026-09-26); look for third-party conversions.
- On a consumer GPU, the 8B VL 2 is an easy size. The 32B models need quantization.
Sources: the model card of karakuri-ai/karakuri-vl-2-8b-thinking-2603; the model list of the karakuri-ai organization on Hugging Face (all as of 2026-09-26).
Our Coverage and Data
Local Model Watch has not published an article on KARAKURI’s models yet. The model list below is collected by code from Hugging Face every day. Part of our publisher index; model families link to their family pages.
At a Glance
| Item | Value |
|---|---|
| Hugging Face organization | karakuri-ai |
| Models on Hugging Face | 10 |
| Articles on its own models | 0 |
| Articles on derived models | 0 |
Recent Models on Hugging Face
The publisher lists 10 models in total on Hugging Face. Below are up to 15 of the most recent generative models (text, image, video, audio and so on), the scope of this site, per organization. Classifiers, feature extractors, research components and repositories without a model card are omitted, and the publisher’s own quantized builds and helper variants are folded into the row of the original model. Parameter counts are left blank for repositories that pack weights into integer types, where Hugging Face’s count is not the real parameter count.
→ Scroll horizontally to see all columns
| Released | Model | Task | Parameters | License | Official variants |
|---|---|---|---|---|---|
| 2026-03-27 | karakuri-ai/karakuri-vl-2-8b-thinking-2603 | vision-language | 8.8B | apache-2.0 |
— |
| 2025-07-03 | karakuri-ai/karakuri-vl-32b-instruct-2507 | vision-language | 33.5B | apache-2.0 |
— |
| 2025-07-03 | karakuri-ai/karakuri-vl-32b-thinking-2507-exp | vision-language | 33.5B | apache-2.0 |
— |
| 2025-01-23 | karakuri-ai/karakuri-lm-32b-thinking-2501-exp | text generation | 32.8B | apache-2.0 |
— |
| 2024-06-18 | karakuri-ai/karakuri-lm-8x7b-instruct-v0.1 | text generation | 46.7B | apache-2.0 |
— |
| 2024-05-22 | karakuri-ai/karakuri-lm-7b-apm-v0.2 | text generation | 7.2B | apache-2.0 |
— |
| 2024-04-25 | karakuri-ai/karakuri-lm-8x7b-chat-v0.1 | text generation | 46.7B | apache-2.0 |
— |
| 2024-04-25 | karakuri-ai/karakuri-lm-7b-apm-v0.1 | text generation | 8.5B | gemma |
— |
| 2024-01-26 | karakuri-ai/karakuri-lm-70b-v0.1 | text generation | 69.2B | other |
— |
| 2024-01-26 | karakuri-ai/karakuri-lm-70b-chat-v0.1 | text generation | 69.2B | other |
— |
From the Hugging Face API. Last changed 2026-09-27 (JST). Collected by code.
Last updated 2026-09-27 (JST). The explanation at the top of this page was written with the help of AI from the primary sources it cites. The tables and lists under “Our Coverage and Data" are assembled by code from our article log and the Hugging Face API.