vect0r18/mirror-jmaxcool-bkn1890-vision-only-v2-426f7520

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 3, 2026Architecture:Transformer Featherless Exclusive Cold

The vect0r18/mirror-jmaxcool-bkn1890-vision-only-v2-426f7520 model is a 35.1 billion parameter vision-only model derived from the BKN1890/albedo-qwen3.6 base. This model specializes in visual processing tasks, focusing exclusively on its visual components. It is designed for applications requiring robust image understanding and analysis, distinguishing itself by its dedicated vision-only profile.

Loading preview...

Model Overview

The vect0r18/mirror-jmaxcool-bkn1890-vision-only-v2-426f7520 is a specialized 35.1 billion parameter model, an "Albedo SN97 scrub candidate" based on the BKN1890/albedo-qwen3.6 architecture. This version is explicitly configured as vision-only, meaning its functionality is restricted to visual processing components (model.visual.*).

Key Characteristics

  • Vision-Only Profile: Unlike general-purpose multimodal models, this variant exclusively utilizes its visual processing capabilities, making it highly focused on image-related tasks.
  • Parameter Count: With 35.1 billion parameters, it offers substantial capacity for complex visual understanding.
  • Context Length: The model supports a context length of 32768 tokens, which can be beneficial for processing detailed visual inputs or sequences.
  • Derived from BKN1890/albedo-qwen3.6: It originates from a specific snapshot of the BKN1890 base model, indicating a targeted development path.
  • Scrubbed Tensors: The model has undergone a "scrubbing" process, with 63 out of 1045 tensors (all vision-related) being selectively modified based on a seed-aware selection, suggesting optimization or fine-tuning for its vision-only role.

Use Cases

This model is particularly well-suited for applications where dedicated visual understanding is paramount, without the need for language generation or other modalities. Potential use cases include:

  • Image Recognition and Classification
  • Object Detection and Segmentation
  • Visual Feature Extraction
  • Image Captioning (when paired with a separate language model)
  • Visual Question Answering (VQA) systems focusing on image interpretation

Finetunes
1 models