AI Models

Platform-admin only. Which model each source runs on, what it costs, and how the fixed reviewer scores what the generator writes. Set a model per source under Settings → Platform (keys openai_model_content_generator, openai_model_content_reviewer); a claude-* value routes through Anthropic with the same budget and logging.

Configured models

Source	Model	Where it comes from

Quality, by generator model

One row per generation run. First-try pass rate is the share of runs where the best candidate cleared the review floor without refinement; attempts is how many drafts were written per run.

Model	Runs	Avg best score	Pass rate	First-try pass	Avg attempts

Cost and latency, by model and source

Model	Source	Calls	Tokens	Cost	Avg latency