PromptVersions
promptversions.com
- Free plan
- Web
What PromptVersions is
PromptVersions is a web workspace for creating, evaluating, and shipping prompts across OpenAI, Anthropic, Google, DeepSeek, xAI, Mistral, and OpenRouter. Saving a prompt creates an immutable version with a commit message, which users can compare or restore. Ordered prompt chains allow later steps to use outputs from earlier ones. Users can define evaluation metrics and test cases, then evaluate a version in a batch or run evaluations whenever it is saved. Versions can be assigned to development, staging, and production environments, tested in A/B experiments, and accessed through an inference API. Workspaces offer owner, editor, and viewer roles, and invitations must be accepted. The service describes signed webhooks for deployments, stopped A/B tests, and evaluation runs. Users connect their own provider API keys, with model usage billed directly by the provider; saved keys are encrypted at rest and are not returned to the browser. The Free plan costs nothing, requires no payment method, and has no charges.
Who it suits
PromptVersions is designed for prompt engineers, and may also suit product managers, writers, and analysts who edit prompts. Its roles and deployment environments can support teams managing prompt changes across development, staging, and production.
Good to know, and what to watch
- Immutable versions can be compared or restored
- Evaluation can run on every save
- Supports seven listed model providers
- Includes development, staging, and production environments
- Free plan requires no payment method
- Provider usage is billed separately by providers
- Users must connect their own API keys
Plans
Every plan and pricePro
Team
Free
Free plan
The verdict
PromptVersions covers prompt version control, evaluation, and deployment workflows, with multiple provider connections and workspace roles. Its free plan has no charges, while paid-plan prices and provider usage costs need to be considered separately.
What the maker publishes
- Bring your own keys
- Users connect their own provider API keys, and provider usage is billed directly by the provider. source
- Evaluations
- Users can define metrics and test cases, then batch-evaluate a version or run evaluations automatically on every save. source
- Intended users
- The maker describes the product as built for prompt engineers and says it supports product managers, writers, and analysts who edit prompts. source
- Key security
- Provider keys are encrypted at rest with Fernet and are write-only in the UI, so they are not returned to the browser after saving. source
- Production workflow
- Prompt versions can be pinned to dev, staging, and prod environments, A/B tested, and called through an inference API. source
- Prompt chains
- Users can connect ordered prompt steps so later steps can use earlier steps’ outputs. source
- Provider integrations
- The service lists OpenAI, Anthropic, Google, DeepSeek, xAI, Mistral, and OpenRouter as supported providers. source
- Purpose
- PromptVersions is a workspace for building, versioning, testing, and shipping LLM prompts across seven major providers. source
- Refund support
- Refund requests are submitted through the in-app Feedback button, and the policy says requests are acknowledged within two business days. source
- Team access
- Workspaces support owner, editor, and viewer roles, with invitations requiring acceptance and access enforced on every route. source
- Version history
- Each prompt save creates an immutable version with a commit message, and users can diff or restore versions. source
- Webhooks
- The site describes signed webhooks for deployments, stopped A/B tests, and evaluation runs. source
Checked against the maker's pages on 3 Oct 2026.