Publications

AI21 Labs’ Summarize API Outperforms OpenAI's Models in Human Evaluation

AI21 Labs’ Summarize API Outperforms OpenAI's Models in Human Evaluation

The performance of AI21 Labs’ Summarize API was evaluated against GPT's Davinci-003 and GPT-3.5-Turbo models, using automated metrics and human evaluation. Summarize API produces shorter summaries with higher pass rates and better faithfulness scores in all experiments utilizing real-world data.