- vendor
- Maxim
- whatItIs
- End-to-end evaluation and observability platform for AI agents spanning prompt experimentation, agent simulation and evals, real-time production tracing, and a built-in LLM gateway (Bifrost) with semantic caching.
- hosting
- both
- pricingModel
- Free 10k logs/mo; Pro $29/seat/mo (100k logs); Business $49/seat/mo (500k logs); Enterprise custom. VPC self-hosted for enterprise.
- modelProviders
- agnostic (OpenAI, Anthropic, Google Gemini, LangChain, LangGraph, CrewAI, Mistral, Bedrock, LiteLLM)
- stateModel
- trace-based logs with session, trace, and span levels
- toolModel
- SDK (Python, TypeScript, Java, Go) + dashboard + Bifrost LLM gateway
- includesBrowser
- false
- includesMemory
- false
- longRunning
- unlimited
- hitlSupport
- true
- observability
- 50+ built-in evaluators, real-time tracing, alerts, custom dashboards, data curation to eval datasets
- maturity
- ga
- launched
- 2025
- notes
- Founded 2023 by ex-Google/Postman engineers; Bifrost gateway adds semantic caching (up to 30% cost reduction). Distinct from LangSmith (no gateway), Langfuse (no experimentation), AgentOps (agent-only). Bundles the Bifrost LLM gateway (see bifrost-maxim).
