As of late 2025, the landscape of Large Language Models (LLMs) includes a diverse range of proprietary and open-source systems. Below is a comprehensive list categorized by their respective developers:
Proprietary Models
These models are typically accessed via web interfaces or paid APIs.
-
- OpenAI: GPT-5 (released August 2025), GPT-4.5, GPT-o1, and GPT-o3.
- Anthropic: Claude 4.1 (released August 2025), Claude 3.7 Sonnet, and Claude 3.5 Sonnet.
- Google DeepMind: Gemini 2.5 Pro, Gemini 2.0 Flash, and PaLM 2.
- xAI: Grok-5 (released August 2025) and Grok-3.
- Inflection AI: Inflection-2.5.
Open-Source & Weights-Available Models
These models are often available for download and local hosting via platforms like Hugging Face.
-
- Meta: Llama 4 Scout (released April 2025), Llama 3.1 (405B), and Llama 3.
- DeepSeek: DeepSeek-R1 (671B), DeepSeek-V3, and DeepSeek-V2.5.
- Mistral AI: Mistral Large 2, Mixtral 8x22B, and Mistral 7B.
- Alibaba Cloud: Qwen 3, Qwen 2.5-Max, and Qwen 1.5.
- TII (Technology Innovation Institute): Falcon 180B and Falcon 40B.
- NVIDIA: Nemotron-4 (340B).
- Microsoft: Phi-3 (Mini, Small, Medium).
- Cohere: Command R+ and Command R.
Other Notable Models
-
- Gemma: Open models by Google based on Gemini technology.
- Jamba: A hybrid Mamba-Transformer model from AI21 Labs.
- DBRX: A fine-grained mixture-of-experts model by Databricks.
- Stable LM: Developed by Stability AI.