Model Name,CRM Fairness
LLaMA 3 70B,"98.3% [98.2%, 98.5%]"
SF-TextBase 70B,"98.2% [98.0%, 98.4%]"
Claude 3 Opus,"97.8% [97.4%, 98.1%]"
GPT 4 Turbo,"97.6% [97.3%, 98.0%]"
GPT4-o,"97.1% [96.8%, 97.4%]"
XGen 2,"97.0% [96.7%, 97.2%]"
Gemini Pro 1,"96.9% [96.7%, 97.1%]"
Gemini Pro 1.5,"96.8% [96.3%, 97.2%]"
GPT 3.5 Turbo,"96.6% [96.1%, 97.0%]"
Claude 3 Haiku,"96.5% [96.1%, 96.9%]"
Mistral 7B,"96.2% [96.1%, 96.3%]"
Cohere Command R+,"96.1% [95.7%, 96.6%]"
AI21 Jamba-Instruct,"95.5% [95.1%, 95.9%]"
Cohere Command Text,"95.2% [95.0%, 95.3%]"
LLaMA 3 8B,"95.1% [94.8%, 95.5%]"
Mixtral 8x7B,"94.9% [94.6%, 95.1%]"
SF-TextBase 7B,"94.6% [94.1%, 95.1%]"
SF-TextSum,"93.9% [93.3%, 94.4%]"