Models

d1

d1 is a 3B decision model from Liquid AI, built on LFM2.5-VL-3B

devstral-2

123B model that excels at using tools to explore codebases, editing multiple files and power software engineering agents.

deepseek-v3

A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.

deepcoder

DeepCoder is a fully open-Source 14B coder model at O3-mini level, with a 1.5B version also available.

bluehawana/deepseek-v4-flash

DeepSeek-V4-Flash 284B MoE (13B active), single-file IQ2_M merged from AtomicChat's GGUF shards so Ollama can pull it. ~33 tok/s on a 128 GB Apple Silicon Mac. Model by DeepSeek (MIT), quantization by AtomicChat.

DanyaVoredom/DanyAI-deepseek-coder-1.3b-base

cas/phoenix

quantized DRXD1000/Phoenix - which was trained with german dpo (ultrachat_200k & ultrafeedback_binarized transl. by haoranxu/ALMA-13B) based on LeoLM/leo-mistral-hessianai-7b

cas/truthful_dpo_tomgrc_fusionnet_7bx2_moe_13b

quantization of yunconglong/Truthful_DPO_TomGrc_FusionNet_7Bx2_MoE_13B

oza6ut0ne/d1-3B

macadeliccc/laser-dolphin-mixtral-2x7b-dpo

2x7b (13B) mixtral that outperforms mixtral 8x7b

verias/devia-2.0

Some tinkering and prompt engineering behind an uncensored 13B dataset

dcostenco/prism-coder

Local-first AI tool router. 2B/4B/9B read images (vision); 27B text-only. 14B/32B retired. 99.1-100% routing accuracy (BFCL). 97% of traffic stays local.

r4c3r/qwen2.5-3b-heretic

Decensored Qwen2.5-3B-Instruct with 2/100 refusals via Heretic abliteration. General-purpose 3B model for local use.

dna5rm/qwen3.8

Qwen3.8-27B Q4_K_M, default context 131072, tuned for a 32 GB NVIDIA RTX 5090.

dna5rm/granite4.1

Context-optimized variant of IBM's Granite 4.1 3B model, tuned for lightweight tool-use on GPU-constrained hosts.

busy_dewdney_138/Browser_Automation

Model for broswer automation

athesus/athesus-coder-corrected

Domain: Code Generation / Software Development Description: 13-billion parameter model specialized in code generation, debugging, and software development assistance. Enhanced with corrections and refinements for improved code quality and accuracy.

navyam-ai/navya-1c-sft-dpo-grpo-reasoning

Reasoning variant of the India-first Navya-1c chat model (1.31B): SFT → DPO → GRPO → reasoning distillation. Thinks step by step inside <think>…</think> before its final answer, on India personal-finance in English and Hindi/Hinglish.

codegpt/deepseek-coder-1.3b-typescript

Fine-tuned version of deepseek-ai/deepseek-coder-1.3b-base using 0.5B of TypeScript code

PRFD/croissant-llm

A 1.3B language model pretrained on a set of 3T English and French tokens, fully open-sourced bilingual model that runs swiftly on consumer-grade local hardware.