Vireqo-27VT-260818

Vita0818/Vireqo-27VT-260818

Vireqo-27VT-260818 is a image-text-to-text model from Vita0818. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for llama.cpp. The card lists the license as apache-2.0.

Vireqo-27VT-260818 is the independent Vision + bounded-Thinking product. It reuses the unchanged standard Vireqo language GGUF and the existing shared Bonsai Q8 projector, then applies the validated Think512-Concise L…

Downloads · 30 days

13% of all-time downloads

All-time downloads

305

Public

Repo size

5.5 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.gguf5.5 GB · 100%

At a glance

Task: Image-Text-to-Text
Library: llama.cpp
License: apache-2.0
Access: Public
Created: Aug 18, 2026
Updated: Aug 18, 2026
SHA: 7961b6b6

Base models

Model card

From the Hugging Face model README

Vireqo-27VT-260818

Vireqo-27VT-260818 is the independent Vision + bounded-Thinking product. It reuses the unchanged standard Vireqo language GGUF and the existing shared Bonsai Q8 projector, then applies the validated Think512-Concise LM Studio configuration.

Required files

File	SHA-256
`Vireqo-27VT-260818.gguf`	`54511bc32a461c6558d0c74278f4e75357cc9dc9fc2b39fbd5b52940ddc031d5`
`mmproj-Vireqo-27VT-260818-Q8_0.gguf`	`eb561d41a7bbeb0fcf04883c8af11078ef6cae0a66862a0b68443cfca495269d`

Both are local symbolic-link aliases to the already stored external-drive files. No additional main or projector weight copy was created.

Thinking preset

Reasoning Budget 512, maximum response 768, temperature 0, repeat penalty 1.08, context 2048, parallel 1.

思考预算已到。复核已有结论后，只输出最终答案；禁止重复或展示思考过程。

Validation

Text Thinking inherited the standard T 3/3 PASS. The chart-image Thinking probe also passed: the model reasoned over the image and returned 2024=40, 2025=70, 2026=55, highest 2025, and difference 15, with no internal marker leak. See thinking-vision-validation.md.

Limitations

Unrestricted Thinking is not accepted.
Natural-image benchmarks, video, dense OCR, and high-stakes vision are not accepted.
Think512-Concise favors short finals.
This is a configuration bundle, not newly trained VLM reasoning weights.

See LM-STUDIO-使用指南.md, TECHNICAL_README.md, and bundle-provenance.json.

Task: Image-Text-to-Text
Library: llama.cpp
License: apache-2.0
Languages: zh, en
Created: Aug 18, 2026
Updated: Aug 18, 2026