Beam 1.0 501B A23B

Qwen 3.8 27B Uncensored Fast by Heabsy

Heabsy's NVFP4 build of Qwen3.8-27B by Alibaba Cloud, with refusal removal based on huihui-ai's Huihui-Qwen3.8-27B-abliterated. Quantized and calibrated by Heabsy, with unpublished weights under the Heabsy Model License. Supports text and image input, tools, and optional reasoning capped at 16,384 tokens within the output budget. This named build is separate from Qwen 3.8 27B Uncensored.

Moondream 3 Preview

Moondream 3 Preview is the original 9B mixture-of-experts vision-language checkpoint with 2B active parameters, for visual question answering, captioning, OCR, and visual reasoning. It accepts text or one image with optional reasoning, a 32,000-token context, and up to 4,096 output tokens.

Moondream 3.1 9B A2B

Moondream 3.1 is the successor to Moondream 3 Preview: a distinct 9B mixture-of-experts vision-language checkpoint with 2B active parameters, for visual question answering, captioning, OCR, and visual reasoning. It accepts text or one image with optional reasoning, a 32,000-token context, and up to 4,096 output tokens.

Orion 26B A4B v1.1

TheDrummer's Gemma 4 26B A4B fine-tune for creative writing, roleplay, and expressive dialogue. Supports optional thinking with a 65,536-token context window and up to 8,192 output tokens. Text input only.

Gemma SEA-LION v4 27B

AI Singapore’s Gemma-based model for text chat in Southeast Asian languages, with a 128K context window. Open weights under the Gemma license; this route supports text input only.

Qwen SEA-LION v4 32B

AI Singapore’s Qwen-based multilingual model, adapted for Southeast Asian languages and cultural contexts. Supports text chat and tool calling with a 128K context window.