ImprovePrompt vs ProofHound

ImprovePrompt has no write-up on Specifiction yet; its figures are above.

ProofHound is a prompt-optimization platform for LLM classification work, available through an API, self-hosted deployment, or the web. It examines error cases, iterates prompts, validates outputs, and provides controlled deployment and rollback. Dataset uploads accept CSV, TSV, JSONL, JSON arrays, and ZIP files, with field mapping in the interface. Optimization can target overall accuracy or category measures such as recall for high-risk categories and precision for error-prone classes. Immutable prompt versions retain variable configurations, output rules, and differences between revisions. Experiment records include datasets, model configurations, sample judgments, and overall and category metrics. Release controls include gray traffic, A/B testing, full rollout, and one-click rollback. Connections include the web UI, webhooks, API tokens, and MCP. Customers supply their own model provider, and ProofHound says model calls are not charged. A free plan is listed at free, while self-hosted open source is also free.