What is the difference between Iris and LangSmith?

Iris is an MCP-native agent eval tool that needs no SDK. Add the config block, restart your client, and every session lists Iris's tools on connect. Iris never intercepts: it runs when your agent calls one of its tools, when a host hook or `iris-eval ingest` hands it a trace, or when you POST one to its HTTP API. LangSmith is LangChain's cloud-first observability platform with @traceable decorators and broad SDK support. Iris offers simple self-hosting with a single SQLite file, while LangSmith

Is Iris better than LangSmith for MCP agent evaluation?

For MCP-compatible agents, Iris provides protocol-native integration with zero SDK overhead and unlimited data retention. LangSmith requires @traceable decorators but offers LLM-as-Judge evaluation, human review workflows, and deep LangChain ecosystem integration. The best choice depends on whether you prioritize zero-code MCP simplicity or broad framework-locked tracing.

Comparison

Iris vs LangSmith

MCP-native, zero-code observability vs SDK-based instrumentation with cloud-first architecture. Two different philosophies for monitoring your AI agents.

TL;DR

Iris is an MCP server your agent discovers and uses automatically — zero code changes, zero SDK imports, one SQLite file for storage. LangSmith is LangChain's full-featured observability and evaluation platform with SDK integrations for multiple frameworks, LLM-as-Judge evaluation, and enterprise-grade infrastructure. If you're building with MCP-compatible agents and want the simplest possible setup with no vendor lock-in, Iris gets you there in 60 seconds. If you need advanced evaluation workflows, auto-clustering, or enterprise compliance today, LangSmith is the more established platform.

For background on agent evaluation methodology, see our agent eval guide.

Feature Comparison

Side by side.

Feature	Iris	LangSmith
Integration method	MCP config (zero code)	SDK imports + @traceable decorators
Self-hosting complexity	Single SQLite file	Enterprise-only, license key required
Performance overhead	Zero (no SDK in hot path)	Async tracing via SDK in your process
Eval rules	20 built-in heuristic rules + 9 custom-rule types (in-process, no provider call)	LLM-as-Judge + human review workflows
Cost tracking	Per-trace USD cost	Token + latency per trace and tool call
MCP support	Protocol-native (IS an MCP server)	LangSmith MCP server (fetch runs, projects, datasets, experiments; no tool runs an eval) plus MCP/A2A deployment support
License	MIT (fully permissive)	Proprietary platform (SDK is MIT)
Pricing	Free + Cloud waitlist	Free tier (5k traces/mo), Plus $39/seat/mo, Enterprise custom
Dashboard	Real-time dark-mode UI	Auto-clustering, pattern detection, custom dashboards
Framework support	Any MCP-compatible agent	LangChain, OpenAI, Anthropic, Vercel AI, LlamaIndex + more
Data retention	Unlimited (your SQLite, your storage)	14 days (free) / 400 days (paid)
Enterprise features	Not offered — self-hosted; no compliance certification claimed before it is held	SSO, BYOC, SOC 2, dedicated support

Decision Guide

Which one fits your stack?

When to choose Iris

You're building with MCP-compatible agents (Claude Desktop, Cursor, Windsurf)
You want zero-code integration — no SDK imports, no decorators
You want simple self-hosting — one binary, one SQLite file, no license key
You want fully permissive MIT licensing with no proprietary lock-in
You want unlimited data retention without per-trace pricing

When to choose LangSmith

You need LLM-as-Judge evaluation with human review workflows
You need enterprise compliance and dedicated support today
You're heavily invested in the LangChain ecosystem
You need auto-clustering and pattern detection across traces
You need broad SDK support across many non-MCP frameworks

Last verified: September 2026. This comparison is based on publicly available documentation and may not reflect recent changes to LangSmith. We aim to keep this page accurate and fair.

See something outdated or incorrect? Report an inaccuracy — we review and update within 48 hours.

Iris vs LangSmith

Side by side.

Which one fits your stack?

When to choose Iris

When to choose LangSmith

Ready to see what your agents are doing?