Navigate to the settings page from the top right menu. Then select API keys under organization settings side panel. You can click the generate button to create your patched_api_key.

This key can be used directly with our OpenAI compatible chat completions endpoint (https://patchwork.patched.codes/v1).

You can also use the patched_api_key with the Patchwork CLI to run patchflows locally.

​
Automated Evaluation

You can use our evaluation endpoint (https://patchwork.patched.codes/evaluate/v1) for testing and debugging your requests. When using the evaluation endpoint only those responses that pass our automated evaluation checking are returned, otherwise the reponse is empty.

You can read more about how we do automated evaluations on our blog.

​
Optimized Inference

You can use our optimized inference endpoint (https://patchwork.patched.codes/optimize/v1) for better accuracy and performance.

We use a number of techniques during inference to improve the performance of models across diverse tasks without the need to do fine-tuning. You can read about them on our blog.