deepseek-ai/DeepSeek-V3.2

Text Generation • 685B • Updated Dec 1, 2025 • 8.1M • • 1.6k

Note My favorite model of all time.

google/gemma-4-31B-it

Image-Text-to-Text • 31B • Updated Jul 20 • 6.81M • • 4.34k

Note My go-to model for a variety of tasks, like conversations / precise translation / roleplay. Really good vision and understands nuance and finer patterns that 26B-A4B doesn't pick up on.

google/gemma-4-26B-A4B-it

Image-Text-to-Text • 26B • Updated Jul 20 • 5.37M • • 1.75k

Note A fantastic model for OCR tasks! Also usable for translations.

google/embeddinggemma-2

Feature Extraction • 0.7B • Updated 15 days ago • 492k • 3.31k

Note My go-to embedding model.

Qwen/Qwen3.8-27B

Image-Text-to-Text • 28B • Updated 5 days ago • 7.23M • • 19.6k

Note While it's a tad slow for programming, it's a great planner and really useful for complex problems that Qwen3.6-35B-A3B can't solve due to MoE sparsity. Also nice for tool calling and websearch.

Qwen/Qwen3.6-27B

Image-Text-to-Text • 28B • Updated 5 days ago • 2.39M • • 2.35k

Note Compared to Qwen3.8-27B, I like the unique voice it has and doesn't have the claudisms.

Qwen/Qwen3.6-35B-A3B

Image-Text-to-Text • 36B • Updated 5 days ago • 4.11M • • 3.11k

Note My go-to programming agent. Very fast and very capable.

kmouratidis/Magistral-Small-2507-Rebased-Vision

Image-Text-to-Text • 24B • Updated Jul 27, 2025 • 46 • • 4

Note The model that gave me the most wholesome roleplay I've had locally.

Delta-Vector/Rei-24B-KTO

Text Generation • 24B • Updated Nov 10, 2025 • 63 • • 19

Note My most used RP model, great at a wide range of topics.

Delta-Vector/Rei-V3-KTO-12B

12B • Updated May 4, 2025 • 24 • 23

Note Great for SFW RPs, has a soft spot in my heart for being the first RP model I've ever used.

Delta-Vector/Hamanasu-Magnum-4B

5B • Updated Apr 7, 2025 • 15 • 5

Note My first RP model on my laptop with only 8GB RAM and no VRAM.

tencent/Hy-MT2-1.8B

Translation • 2B • Updated May 26 • 23.2k • • 1.28k

Note A very tiny and efficient translation model, runnable on 8GB RAM with no VRAM.