➡️ To provide a clearer picture of model performance within an organization, I'll include Elo Rankings :
🔴 Bottom Tier → 🟢 Top Tier
Smaller LLM models are not necessarily older or inferior versions of larger models. They can be designed to run efficiently on lower-end machine specs or for specific tasks, such as text classification or sentiment analysis. These models may use different architectures or techniques to achieve better performance on certain hardware or applications.
➡️ You can find an overall leaderboard at chat.lmsys.org.
Important
If you're concerned about privacy issues related to cloud based LLM providers or you want to experiment with chatbots check out this practical guide on how to set up and run your own model on your local machine.
- Yi-Large (Proprietary) 🟢
- Claude 3.5 Sonnet (Proprietary) 🟢
- Claude 3 Opus (Proprietary)
- Claude 3 Sonnet (Proprietary)
- Claude 3 Haiku (Proprietary)
- Claude 2.1 (Proprietary)
- Claude 2 (Proprietary)
- Claude Instant 1.2 (Proprietary) 🔴
- Gemini Advanced (Proprietary)
- Gemini Flash (Proprietary)
- Gemini 1.5 Pro (Proprietary) 🟢
- Gemini Pro (API) (Proprietary)
- Gemini Pro (Proprietary)
- PaLM-chat-Bison-001 (Proprietary)
- Mistral Large (Proprietary) 🟢
- Mistral Medium (Proprietary)
- Mistral-Next (Proprietary) 🔴
- MPT-30B-chat (Proprietary) 🟢
- MPT-7B-chat (Proprietary) 🔴
- GPT-4o (Proprietary) 🟢
- GPT-4 Turbo (Proprietary)
- GPT-4 (Proprietary)
- GPT-3.5 Turbo (Proprietary) 🔴
- pplx-70b-online (Proprietary) 🟢
- pplx-7b-online (Proprietary) 🔴
Yi-1.5 (upgraded version of Yi)
- Yi-1.5-34B-Chat (Open) 🟢
- Yi-1.5-9B-Chat (Open)
- Yi-1.5-6B-Chat (Open)
Yi
- Yi-34B-Chat (Open)
- Yi-6B-Chat (Open) 🔴
Yi VL (Vision)
- Qwen-Max (Proprietary)
Qwen2
- Qwen2-72B-instruct (Open) 🟢
- Qwen2-7B-instruct (Open)
- Qwen2-1.5B-instruct (Open)
- Qwen2-0.5B-instruct (Open)
Qwen1.5 (the improved version of Qwen)
- Qwen1.5-110B-Chat (Open)
- Qwen1.5-72B-Chat (Open)
- Qwen1.5-32B-Chat (Open)
- Qwen1.5-14B-Chat (Open)
- Qwen1.5-7B-Chat (Open)
- Qwen1.5-4B-Chat (Open)
- Qwen1.5-0.5B-Chat (Open)
Qwen
- Qwen-72B-Chat (Open)
- Qwen-14B-Chat (Open)
- Qwen-7B-Chat (Open)
- Qwen-1.8B-Chat (Open) 🔴
CodeQwen
- CodeQwen1.5-7B-Chat (Open) 🟢
Qwen
- dolphin-2.9.2-qwen2-7b (Open)
- dolphin-2.9.2-qwen2-72b (Open)
Mixtral
- dolphin-2.2.1-mistral-7b (Open) 🔴
- dolphin-2.5-mixtral-8x7b (Open)
- dolphin-2.9.1-mixtral-1x22b (Open)
Llama
- dolphin-2.9-llama3-8b (Open)
- dolphin-2.9.1-llama-3-70b (Open) 🟢
Phi
Aya 23
- aya-23-35B (Open)
- aya-23-8B (Open) 🔴
Command R Plus
- Command R+ (Open) 🟢
Command R
- Command R (Open)
- DBRX (open) 🟢
LLM
- DeepSeek-LLM-67b-Chat (Open) 🟢
- DeepSeek-LLM-7b-Chat (Open) 🔴
Coder
- DeepSeek-Coder-V2-Instruct (Open) 🟢
- DeepSeek-Coder-V2-Lite-Instruct (Open)
- DeepSeek-Coder-33B-instruct (Open)
- DeepSeek-Coder-6.7B-instruct (Open)
- DeepSeek-Coder-7B-instruct-v1.5 (Open)
- DeepSeek-Coder-1.3B-instruct (Open) 🔴
VL (Vision)
- DeepSeek-vl-7b-Chat (Open)
- DeepSeek-vl-1.3b-Chat (Open)
Math
MoE
Gemma
PaliGemma (Vision)
CodeGemma (Coding)
Zephyr ORPO
- Zephyr-orpo-141b-A35b-v0.1 (Open) 🟢
Zephyr 7B
- zephyr-7b-alpha (Open) 🔴
- zephyr-7b-beta (Open)
- zephyr-7b-gemma-v0.1 (Open)
Starchat (Coding)
- Starchat2-15b-v0.1 (Open)
- Vicuna-33B (open) 🟢
- Vicuna-13B (open)
- Vicuna-7B (open)
- FastChat-T5-3B (open) 🔴
- Llama-3-70b-Instruct (Open) 🟢
- Llama-3-8b-Instruct (Open)
- Llama-2-70b-chat (Open)
- Llama-2-13b-chat (Open)
- Llama-2-7b-chat (Open) 🔴
Phi-3
- Phi-3-vision-128k-instruct (Open)
- Phi-3-medium-128k-instruct (Open) 🟢
- Phi-3-small-128k-instruct (Open)
- Phi-3-mini-128k-instruct (Open)
Phi-1
Mixtral (MoE)
- Mixtral-8x7b-Instruct-v0.1 (Open)
- Mixtral-8x22b-Instruct-v0.1 (Open) 🟢
Mistral
- Mistral-7B-Instruct-v0.1 (Open) 🔴
- Mistral-7B-Instruct-v0.2 (Open)
- Mistral-7B-Instruct-v0.3 (Open)
Codestral (Coding)
- Codestral-22B-v0.1 (Open)