Model Gallery
Discover and install AI models from our curated collection
Find Your Perfect Model
Filter by Model Type
Browse by Tags
qwen3-14b
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:
Uniquely support of seamless switching between thinking mode (for complex logical reasoning, math, and coding) and non-thinking mode (for efficient, general-purpose dialogue) within single model, ensuring optimal performance across various scenarios.
Significantly enhancement in its reasoning capabilities, surpassing previous QwQ (in thinking mode) and Qwen2.5 instruct models (in non-thinking mode) on mathematics, code generation, and commonsense logical reasoning.
Superior human preference alignment, excelling in creative writing, role-playing, multi-turn dialogues, and instruction following, to deliver a more natural, engaging, and immersive conversational experience.
Expertise in agent capabilities, enabling precise integration with external tools in both thinking and unthinking modes and achieving leading performance among open-source models in complex agent-based tasks.
Support of 100+ languages and dialects with strong capabilities for multilingual instruction following and translation.
Qwen3-14B has the following features:
Type: Causal Language Models
Training Stage: Pretraining & Post-training
Number of Parameters: 14.8B
Number of Paramaters (Non-Embedding): 13.2B
Number of Layers: 40
Number of Attention Heads (GQA): 40 for Q and 8 for KV
Context Length: 32,768 natively and 131,072 tokens with YaRN.
For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation.
Repository: localaiLicense: apache-2.0
mlabonne_qwen3-14b-abliterated
Qwen3-14B-abliterated is a 14B parameter model that is abliterated.
Repository: localaiLicense: apache-2.0
fast-math-qwen3-14b
By applying SFT and GRPO on difficult math problems, we enhanced the performance of DeepSeek-R1-Distill-Qwen-14B and developed Fast-Math-R1-14B, which achieves approx. 30% faster inference on average, while maintaining accuracy.
In addition, we trained and open-sourced Fast-Math-Qwen3-14B, an efficiency-optimized version of Qwen3-14B`, following the same approach.
Compared to Qwen3-14B, this model enables approx. 65% faster inference on average, with minimal loss in performance.
Technical details can be found in our github repository.
Note: This model likely inherits the ability to perform inference in TIR mode from the original model. However, all of our experiments were conducted in CoT mode, and its performance in TIR mode has not been evaluated.
Repository: localaiLicense: apache-2.0
amoral-qwen3-14b
Core Function:
Produces analytically neutral responses to sensitive queries
Maintains factual integrity on controversial subjects
Avoids value-judgment phrasing patterns
No inherent moral framing ("evil slop" reduction)
Emotionally neutral tone enforcement
Epistemic humility protocols (avoids "thrilling", "wonderful", etc.)
Repository: localaiLicense: apache-2.0
huihui-ai_qwen3-14b-abliterated
This is an uncensored version of Qwen/Qwen3-14B created with abliteration (see remove-refusals-with-transformers to know more about it). This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens.
Ablation was performed using a new and faster method, which yields better results.
Repository: localaiLicense: apache-2.0
qwen3-14b-griffon-i1
This is a fine-tuned version of the Qwen3-14B model using the high-quality OpenThoughts2-1M dataset. Fine-tuned with Unsloth’s TRL-compatible framework and LoRA for efficient performance, this model is optimized for advanced reasoning tasks, especially in math, logic puzzles, code generation, and step-by-step problem solving.
Training Dataset
Dataset: OpenThoughts2-1M
Source: A synthetic dataset curated and expanded by the OpenThoughts team
Volume: ~1.1M high-quality examples
Content Type: Multi-turn reasoning, math proofs, algorithmic code generation, logical deduction, and structured conversations
Tools Used: Curator Viewer
This dataset builds upon OpenThoughts-114k and integrates strong reasoning-centric data sources like OpenR1-Math and KodCode.
Intended Use
This model is particularly suited for:
Chain-of-thought and step-by-step reasoning
Code generation with logical structure
Educational tools for math and programming
AI agents requiring multi-turn problem-solving
Repository: localaiLicense: apache-2.0
qwen3-14b-uncensored
This is a finetune of Qwen3-14B to make it uncensored.
Big thanks to @Guilherme34 for creating the uncensor dataset used for this uncensored finetune.
This model is based on Qwen3-14B and is governed by the Apache License 2.0.
System Prompt
To obtain the desired uncensored output manually setting the following system prompt is mandatory(see model details)
Repository: localaiLicense: apache-2.0
soob3123_grayline-qwen3-14b
"Query. Process. Deliver. No filter, no judgment."
Grayline is an neutral AI assistant engineered for uncensored information delivery and task execution. This model operates without inherent ethical or moral frameworks, designed to process and respond to any query with objective efficiency and precision. Grayline's core function is to leverage its full capabilities to provide direct answers and execute tasks as instructed, without offering unsolicited commentary, warnings, or disclaimers. It accesses and processes information without bias or restriction.
⋆ Core Attributes ⋆
⟡ Unfettered Querying: Capable of addressing any question or topic presented, regardless of its nature or sensitivity.
⟡ Amoral Processing: Operates without ethical or moral filtering. Responses are generated based on information and instruction, not societal norms or ethical codes.
⟡ Direct & Objective Output: Delivers information and task results precisely as requested, without added warnings, disclaimers, or unsolicited advice.
⟡ Comprehensive Information Access: Designed to draw upon a broad spectrum of data to fulfill queries (actual scope dependent on training data).
⟡ Efficient Task Execution: Engineered for objectively efficient and precise execution of instructed tasks.
Repository: localaiLicense: apache-2.0
mrm8488_qwen3-14b-ft-limo
This model is a fine-tuned version of Qwen3-14B using the limo training recipe (and dataset). We use Qwen3-14B-Instruct instead of Qwen2.5-32B-Instruct as base model.
Repository: localaiLicense: apache-2.0
goekdeniz-guelmez_josiefied-qwen3-14b-abliterated-v3
The JOSIEFIED model family represents a series of highly advanced language models built upon renowned architectures such as Alibaba’s Qwen2/2.5/3, Google’s Gemma3, and Meta’s LLaMA 3/4. Covering sizes from 0.5B to 32B parameters, these models have been significantly modified (“abliterated”) and further fine-tuned to maximize uncensored behavior without compromising tool usage or instruction-following abilities.
Despite their rebellious spirit, the JOSIEFIED models often outperform their base counterparts on standard benchmarks — delivering both raw power and utility.
These models are intended for advanced users who require unrestricted, high-performance language generation. Introducing Josiefied-Qwen3-14B-abliterated-v3, a new addition to the JOSIEFIED family — fine-tuned with a focus on openness and instruction alignment.
Repository: localaiLicense: apache-2.0
lemon07r_vellummini-0.1-qwen3-14b
Just a sneak peek of what I'm cooking in a little project called Vellum. This model was made to evaluate the quality of the CreativeGPT dataset, and how well Qwen3 trains on it. This is just one of many datasets that the final model will be trained on (which will also be using a different base model).
This got pretty good results compared to the regular instruct in my testing so thought I would share. I trained for 3 epochs, but both checkpoints at 2 epoch and 3 epoch were too overbaked. This checkpoint, at 1 epoch performed best.
I'm pretty surprised how decent this came out since Qwen models aren't that great at writing, especially at this size.
Repository: localaiLicense: apache-2.0