THE CURIOUS SIDE OF AI
QwenTEXT MODELQwen3-0.6B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
CPU / small GPURead profile →
AIBRAINWORX / MODEL LIBRARY
Find your next
local possibility.
Text, voice, vision, video and very small edge candidates. Understand the model here before choosing where to download it.
56 model profiles. Six dataset guides.
Source-reviewed summaries with original creator credit. Read runtime requirements, exact licenses and test status before downloading. Specialized profiles need their own runtimes; no weights are hosted here. Browse dataset preparation guides →
56 models
QwenTEXT MODELQwenTEXT MODELQwen3-1.7B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
CPU / small GPURead profile →
QwenTEXT MODELQwen3-4B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
CPU / small GPURead profile →
QwenTEXT MODELQwen3-8B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
Mid-range GPU / CPU offloadRead profile →
QwenTEXT MODELQwen3-14B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
Mid-range GPU / CPU offloadRead profile →
QwenTEXT MODELQwen3-32B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
Large GPU / more RAMRead profile →
QwenTEXT MODELQwen3-30B-A3B
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. It offers thinking and non-thinking modes through its supported chat template.
Large GPU / more RAMRead profile →
QwenTEXT MODELQwen3-4B-Instruct-2507
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. This July 2025 Instruct release answers without a separate thinking mode.
CPU / small GPURead profile →
QwenTEXT MODELQwen3-30B-A3B-Instruct-2507
An instruction-tuned Qwen text model for conversation, writing, coding and multilingual tasks. This July 2025 Instruct release answers without a separate thinking mode.
Large GPU / more RAMRead profile →
HuggingFaceTBTEXT MODELSmolLM2-135M-Instruct
A compact instruction-tuned language model from Hugging Face’s SmolLM2 family for lightweight text generation and on-device experiments.
CPU / small GPURead profile →
HuggingFaceTBTEXT MODELSmolLM2-360M-Instruct
A compact instruction-tuned language model from Hugging Face’s SmolLM2 family for lightweight text generation and on-device experiments.
CPU / small GPURead profile →
HuggingFaceTBTEXT MODELSmolLM2-1.7B-Instruct
A compact instruction-tuned language model from Hugging Face’s SmolLM2 family for lightweight text generation and on-device experiments.
CPU / small GPURead profile →
HuggingFaceTBTEXT MODELSmolLM3-3B
A 3B instruction model with optional reasoning, six supported languages and long-context training. Hugging Face publishes its training details and data-mixture information.
CPU / small GPURead profile →
mistralaiTEXT MODELMistral-7B-Instruct-v0.3
The v0.3 instruction release of Mistral 7B, adding its v3 tokenizer and function-calling support.
Mid-range GPU / CPU offloadRead profile →
mistralaiTEXT MODELMistral-Nemo-Instruct-2407
A 12B instruction model jointly developed by Mistral AI and NVIDIA, with multilingual text capabilities and a 128K trained context window.
Mid-range GPU / CPU offloadRead profile →
mistralaiVISIONMistral-Small-3.1-24B-Instruct-2503
A 24B multimodal instruction model supporting text and image understanding, with a 128K context window.
Large GPU / more RAMRead profile →
mistralaiTEXT MODELMixtral-8x7B-Instruct-v0.1
An instruction-tuned mixture-of-experts text model that routes each token through selected experts.
Large GPU / more RAMRead profile →
ibm-graniteTEXT MODELgranite-3.3-2b-instruct
An IBM Granite 3.3 instruction model for language tasks including tool use, retrieval-grounded answers, reasoning and multilingual conversations.
CPU / small GPURead profile →
ibm-graniteTEXT MODELgranite-3.3-8b-instruct
An IBM Granite 3.3 instruction model for language tasks including tool use, retrieval-grounded answers, reasoning and multilingual conversations.
Mid-range GPU / CPU offloadRead profile →
ibm-graniteTEXT MODELgranite-4.0-h-tiny
IBM’s 7B hybrid instruction model combines Mamba-style and attention layers with sparse experts, long context and instruction following.
Mid-range GPU / CPU offloadRead profile →
allenaiTEXT MODELOLMo-2-1124-7B-Instruct
An Allen Institute for AI instruction-tuned OLMo 2 text model, accompanied by open training and research resources.
Mid-range GPU / CPU offloadRead profile →
allenaiTEXT MODELOLMo-2-0325-32B-Instruct
An Allen Institute for AI instruction-tuned OLMo 2 text model, accompanied by open training and research resources.
Large GPU / more RAMRead profile →
deepseek-aiTEXT MODELDeepSeek-R1-Distill-Qwen-1.5B
A Qwen-based distilled reasoning model trained using DeepSeek-R1 outputs for mathematical, coding and reasoning tasks.
CPU / small GPURead profile →
deepseek-aiTEXT MODELDeepSeek-R1-Distill-Qwen-7B
A Qwen-based distilled reasoning model trained using DeepSeek-R1 outputs for mathematical, coding and reasoning tasks.
Mid-range GPU / CPU offloadRead profile →
deepseek-aiTEXT MODELDeepSeek-R1-Distill-Qwen-14B
A Qwen-based distilled reasoning model trained using DeepSeek-R1 outputs for mathematical, coding and reasoning tasks.
Mid-range GPU / CPU offloadRead profile →
deepseek-aiTEXT MODELDeepSeek-R1-Distill-Qwen-32B
A Qwen-based distilled reasoning model trained using DeepSeek-R1 outputs for mathematical, coding and reasoning tasks.
Large GPU / more RAMRead profile →
microsoftTEXT MODELPhi-4-mini-instruct
Microsoft’s compact 3.8B instruction model for text reasoning, language understanding and tool-oriented tasks.
CPU / small GPURead profile →
microsoftTEXT MODELPhi-3-mini-4k-instruct
Microsoft’s 3.8B instruction-tuned Phi-3 Mini release with a 4K context window for compact text-generation workloads.
CPU / small GPURead profile →
QwenTEXT MODELQwen2.5-0.5B-Instruct
An instruction-tuned Qwen2.5 text model for chat, structured answers, coding and multilingual writing.
CPU / small GPURead profile →
QwenTEXT MODELQwen2.5-1.5B-Instruct
An instruction-tuned Qwen2.5 text model for chat, structured answers, coding and multilingual writing.
CPU / small GPURead profile →
QwenTEXT MODELQwen2.5-7B-Instruct
An instruction-tuned Qwen2.5 text model for chat, structured answers, coding and multilingual writing.
Mid-range GPU / CPU offloadRead profile →
QwenTEXT MODELQwen2.5-14B-Instruct
An instruction-tuned Qwen2.5 text model for chat, structured answers, coding and multilingual writing.
Mid-range GPU / CPU offloadRead profile →
QwenTEXT MODELQwen2.5-32B-Instruct
An instruction-tuned Qwen2.5 text model for chat, structured answers, coding and multilingual writing.
Large GPU / more RAMRead profile →
DeepSeekRESEARCH PROFILEDeepSeek-V4.1-Flash-1017
The October 17 update improves the publisher’s reported coding, reasoning, agent and visual evaluations while reducing output length and task time.
Large server deploymentRead profile →
QwenRESEARCH PROFILEQwen3.8-Flash-Next
A multimodal sparse model combining linear and sparse attention, mixture-of-experts routing, n-gram embeddings and multi-token prediction.
Large RAM / specialized offload engineRead profile →
QwenRESEARCH PROFILEQwen3.8-27B
A dense multimodal model using a hybrid of Gated DeltaNet and attention layers, with multi-token prediction and selectable reasoning behavior.
Large GPU / quantization and RAM offloadRead profile →
Moonshot AIRESEARCH PROFILEKimi-K3
A native multimodal mixture-of-experts model for long-running coding, research and knowledge work, using Kimi Delta Attention and Attention Residuals.
Large server deploymentRead profile →
NVIDIARESEARCH PROFILECosmos3-H-Dreams
An action-conditioned surgical world model that generates video from robot state and action inputs, rather than serving as a general chat assistant.
Specialized research deploymentRead profile →
GoogleTEXT MODELgemma-3-270m-it
A very small instruction-tuned Gemma text model for focused tasks on resource-constrained hardware.
Very small model / edge candidate; runtime validation requiredRead profile →
GoogleTEXT MODELgemma-3-1b-it
Google’s compact instruction-tuned Gemma 3 text model for local and on-device applications.
Very small model / edge candidate; runtime validation requiredRead profile →
GoogleTEXT MODELfunctiongemma-270m-it
A tiny Gemma model specialized for translating requests into function-call structures.
Very small model / edge candidate; runtime validation requiredRead profile →
GoogleEMBEDDINGSembeddinggemma-300m
A compact multilingual embedding model for semantic retrieval and similarity, rather than answer generation.
Very small model / edge candidate; runtime validation requiredRead profile →
Hugging FaceVISIONSmolVLM-256M-Instruct
A small vision-language instruction model for image descriptions and questions with limited compute.
Very small model / edge candidate; runtime validation requiredRead profile →
Hugging FaceVISIONSmolVLM-500M-Instruct
A compact vision-language model for image-grounded conversation and lightweight visual experiments.
Very small model / edge candidate; runtime validation requiredRead profile →
hexgradVOICE / TTSKokoro-82M
A lightweight text-to-speech model with Apache-licensed weights for local speech synthesis.
Very small model / edge candidate; runtime validation requiredRead profile →
QwenVOICE / TTSQwen3-TTS-12Hz-0.6B-Base
The small Qwen3-TTS Base checkpoint supports multilingual speech synthesis from reference voice audio.
Very small model / edge candidate; runtime validation requiredRead profile →
QwenVOICE / TTSQwen3-TTS-12Hz-0.6B-CustomVoice
A small Qwen speech checkpoint for controllable synthesis with supported preset speaker identities.
Very small model / edge candidate; runtime validation requiredRead profile →
KyutaiVOICE / TTSpocket-tts
A small CPU-oriented speech synthesis family for local voice applications.
Very small model / edge candidate; runtime validation requiredRead profile →
KittenMLVOICE / TTSkitten-tts-mini-0.8
A compact Kitten speech model distributed for lightweight local text-to-speech generation.
Very small model / edge candidate; runtime validation requiredRead profile →
KittenMLVOICE / TTSkitten-tts-nano-0.8-fp32
A very small Kitten speech model aimed at local synthesis without a large GPU.
Very small model / edge candidate; runtime validation requiredRead profile →
Sentence TransformersEMBEDDINGSall-MiniLM-L6-v2
A small sentence embedding model for English semantic search, clustering and similarity.
Very small model / edge candidate; runtime validation requiredRead profile →
BAAIEMBEDDINGSbge-small-en-v1.5
A small English embedding checkpoint for retrieval-oriented text representations.
Very small model / edge candidate; runtime validation requiredRead profile →
OpenAISPEECH RECOGNITIONwhisper-tiny
A tiny multilingual Whisper checkpoint for automatic speech recognition and speech translation.
Very small model / edge candidate; runtime validation requiredRead profile →
GoogleVISIONmobilenet_v2_1.0_224
A compact MobileNet V2 image classifier pretrained for 224-by-224 image input.
Very small model / edge candidate; runtime validation requiredRead profile →
Wan-AIVIDEO / AVATARWan 2.1 / small video
A compact text-to-video model for short creative scenes and effects experiments.
Consumer GPU candidate; publisher lists 8.19 GB VRAMRead profile →
Wan-AIVIDEO / AVATARWan 2.2 / speech-to-video
An audio-driven video model combining a reference image, speech and optional scene direction.
Large GPU workflow; official example lists 80 GB VRAMRead profile →
No matching profiles. Try a creator name or “CPU”.
What these profiles tell you
Model-card capabilities and context limits describe the original checkpoint. Your actual fit and speed depend on quantization, runtime, RAM, VRAM and context. A model’s license and its creator stay visible; benchmark claims are attributed to their publisher.
Set up AiBrainWorX Model Launcher →