Loading tool details…
Development platform for production LLM apps: prompt engineering, evaluation, retrieval, and workflow orchestration in one workspace.
Development platform for production LLM apps: prompt engineering, evaluation, retrieval, and workflow orchestration in one workspace.
Vellum is a developer platform that bundles the full lifecycle of LLM application development — prompt iteration, evaluation, retrieval-augmented generation, workflow orchestration, and observability — into a single product. The Prompts module is a side-by-side playground where teams compare GPT, Claude, Gemini, and open-weight models across versioned datasets with shared variables and template syntax.
A drag-and-drop canvas for building multi-step LLM pipelines with branching logic, conditional routing, tool use, and RAG integration. Teams can chain prompts, API calls, and code execution nodes without writing orchestration boilerplate. The editor supports real-time collaboration and version history, making it practical for teams iterating on complex pipelines. Each node can be individually tested and debugged, and the visual representation helps non-engineering stakeholders understand and contribute to pipeline design.
Provides automated evaluation pipelines that let teams define custom scoring functions, use LLM-as-judge evaluators, and run side-by-side comparisons of prompt variants. Regression tests can be triggered on every prompt change to catch quality degradations before deployment. The framework tracks quality metrics over time, giving teams quantitative evidence for prompt optimization decisions. This evaluation-driven approach reduces the risk of shipping degraded outputs and provides an audit trail of quality measurements across releases.
Supports 50+ LLMs from providers including OpenAI, Anthropic, Google, Cohere, Meta, and various open-source models through a unified API layer. Teams can swap models at the configuration level without code changes, enabling cost optimization and risk mitigation against provider outages. This also simplifies benchmarking new models against existing production prompts, allowing teams to evaluate whether newer or cheaper models meet their quality thresholds before committing to a migration.
Handles the full retrieval-augmented generation workflow: document ingestion from multiple formats, configurable chunking strategies, embedding generation, and semantic search indexing. Teams can build and iterate on knowledge-base-powered applications directly within the platform rather than stitching together separate vector databases and processing pipelines. The integrated approach means retrieval parameters can be tuned alongside prompt parameters in a unified evaluation loop, leading to faster iteration on RAG quality.
Offers versioned API endpoints with zero-downtime rollouts, A/B testing for prompt variants, and real-time dashboards tracking latency, cost, error rates, and output quality. Rollback capabilities let teams instantly revert to previous prompt or workflow versions if issues arise in production, reducing the blast radius of changes to LLM-powered features. The monitoring layer integrates with the evaluation framework, so teams can set alerts on quality metric thresholds and catch production degradation early.
Starting around $500/mo
Custom
Custom
Ready to get started with Vellum?
View Pricing Options →We believe in transparent reviews. Here's what Vellum doesn't handle well:
In early 2026, Vellum expanded its model gateway to support over 50 LLMs including the latest models from major providers. The platform introduced enhanced workflow editor features with improved real-time collaboration, more granular evaluation metrics for regression testing, and expanded RAG pipeline capabilities with additional document format support. Enterprise customers gained access to improved audit trail reporting and more flexible role-based access control configurations.
AI Evaluation
Braintrust review 2026: unified evals, prompt workbench, observability, and AI proxy used by OpenAI, Stripe, Notion — features, pricing, pros, cons.
LLM Observability
Open-source LLM engineering platform that captures traces, runs evaluations, and helps teams analyze cost, latency, and quality across production AI apps.
AI Monitoring
Open-source LLM observability and gateway platform offering one-line integration, request logging, prompt management, evals, and a multi-provider proxy.
AI Gateway
Portkey is an AI gateway for routing, observing, securing, and governing production traffic across model providers, agents, and MCP tools. It centralizes fallbacks, guardrails, prompt operations, cost tracking, and access policy for engineering teams.
Enterprise AI Agent Platform
Enterprise agent factory for governed low-code workflows
Pricing and features change often. If anything on this Vellum page is wrong or missing, tell us and include a link to the source. We review every correction.