Mar 21, 2026
- Model hubs (free downloads/open weights)
- Hugging Face Hub: https://huggingface.co/models
- ModelScope (Alibaba): https://modelscope.cn/models
- Kaggle Models: https://www.kaggle.com/models
- GitHub model releases: https://github.com (search by model/org name)
- Leaderboards and finders
- Hugging Face Open LLM Leaderboard: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
- LMSYS (Chatbot Arena): https://lmsys.org/
- Papers with Code (SOTA by task): https://paperswithcode.com/sota
- Popular open-weight models (search/download on the hubs above)
- LLMs: Meta Llama 3.1 (https://huggingface.co/meta-llama), Mistral/Mixtral (https://huggingface.co/mistralai), Qwen2/Qwen2.5 (https://huggingface.co/Qwen), Gemma 2 (https://huggingface.co/google), Phi-3 (https://huggingface.co/microsoft), DeepSeek (https://huggingface.co/deepseek-ai)
- Multimodal/Vision-Language: LLaVA (https://huggingface.co/liuhaotian), Qwen2-VL (https://huggingface.co/Qwen), InternVL (https://huggingface.co/OpenGVLab), Idefics2 (https://huggingface.co/HuggingFaceM4)
- Image generation: Stable Diffusion 1.5/SDXL (https://huggingface.co/stabilityai), FLUX.1 (https://huggingface.co/black-forest-labs)
- Speech/audio: Whisper (https://huggingface.co/openai/whisper-large-v3), Bark (https://huggingface.co/suno), MusicGen (https://huggingface.co/facebook)
- Free APIs and playgrounds (rate-limited/free tiers)
- Hugging Face Spaces/Inferences (many models have free demos): https://huggingface.co/spaces
- Groq Playground/API (Llama 3.x, Mixtral, Gemma 2): https://console.groq.com/playground
- Together AI Playground: https://api.together.xyz/playground
- Fireworks.ai Playground: https://app.fireworks.ai/
- OpenRouter Playground (rotating free community models): https://openrouter.ai/playground
- Run locally (free)
- Ollama model library (one-line downloads, GGUF): https://ollama.com/library
- LM Studio (desktop runner): https://lmstudio.ai
- llama.cpp (CPU/GPU quantized): https://github.com/ggerganov/llama.cpp
- Text Generation WebUI / Open WebUI: https://github.com/oobabooga/text-generation-webui and https://github.com/open-webui/open-webui
- For image: Automatic1111 / ComfyUI (Stable Diffusion): https://github.com/AUTOMATIC1111/stable-diffusion-webui and https://github.com/comfyanonymous/ComfyUI
Notes:
- Always check the model’s license (some are research-only or have commercial restrictions).
- Use quantized formats (e.g., GGUF) for low-end hardware; many hubs provide them.
Os melhores lugares para encontrar modelos de IA de ponta gratuitos em 2026 incluem a CometAPI, que oferece um catálogo unificado de modelos com uma chave de API gratuita e créditos de teste. O caminho “gratuito” mais útil depende do seu objetivo: chat e redação do dia a dia, uso multimodal como imagens ou testes de API em muitos modelos de fronteira. A CometAPI destaca-se para desenvolvedores porque oferece uma única API para 500+ modelos, um Playground interativo, créditos de teste para novos usuários e ferramentas de comparação integradas para testar fluxos de trabalho de geração de imagens e vídeos.