Burr (DAGWorks) vs LangGraph Persistence
Burr (DAGWorks) (agent frameworks (no first-party memory layer)) vs LangGraph Persistence (framework-embedded memory): cross-category comparison covering architecture, taxonomy, license, pricing, and direct edges.
How they relate
- Burr (DAGWorks) competes with LangGraph Persistence — lightweight alternative to LangGraph.
Cost & capability
| Burr (DAGWorks) | LangGraph Persistence | |
|---|---|---|
| Capability band | competent | competent |
| Capability composite | 55 | 72 |
| Cost tier | free | free |
| $/Mtok input | 0 | 0 |
| $/Mtok output | 0 | 0 |
| Use cases | Scoped Agentic, Long Running Session | Long Running Session, Multi Agent Coordination, Scoped Agentic, Memory Augmented Chat |
Where they differ (17)
Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.
| Burr (DAGWorks) | LangGraph Persistence | |
|---|---|---|
| Capability composite | 55 | 72 |
| Use cases | Scoped Agentic, Long Running Session | Long Running Session, Multi Agent Coordination, Scoped Agentic, Memory Augmented Chat |
| Section | Agent frameworks (no first-party memory layer) | Framework-embedded memory |
| Tier | T3 | T2 |
| Type | State-machine agent framework | Checkpoints + Store System |
| Created | 2024 | 2024-01 (LangGraph v0.1 released Jan 2024; long-term memory persistence added Oct 2024) |
| Latest release | v0.43.0 (2026-08-25) — apache-burr | langgraph v1.2.16 (2026-10-16) |
| License | Apache-2.0 | MIT |
| GitHub | apache/burr (formerly DAGWorks-Inc/burr) — ~2.6k stars, last commit 2026-08 | langchain-ai/langgraph — ~43k stars, last commit 2026-10 |
| Pricing | free (OSS) | OSS free (MIT); LangSmith Developer $0 (5k traces/mo, 1 seat); Plus $39/seat/mo (1 free small serverless deployment); deployments metered at $1/LSU by compute+memory; Enterprise custom |
| Deployment | Self-hosted only — library + self-hosted tracking UI | Both (LangGraph OSS self-hosted; LangSmith Deployment cloud managed or self-hosted) |
| API surface | searched not found | SDK: Python, JS/TS |
| MCP | searched not found | yes — Agent Server exposes deployed graphs as MCP tools at /mcp (streamable HTTP); MCP client via langchain[mcp] adapter |
| A2A | searched not found | supported — LangGraph A2A wrappers |
| OpenTelemetry | searched not found | first-class — LangSmith + OTel |
| Optimised for | production-grade state-machine LLM workflows with observability | durable graph-state checkpointing + Store System for cross-thread memory |
| Anti-fit | searched not found | not for non-LangGraph workflows |
At a glance
| Burr (DAGWorks) | LangGraph Persistence | |
|---|---|---|
| Section | Agent frameworks (no first-party memory layer) | Framework-embedded memory |
| Tier | T3 | T2 |
| Type | State-machine agent framework | Checkpoints + Store System |
| Created | 2024 | 2024-01 (LangGraph v0.1 released Jan 2024; long-term memory persistence added Oct 2024) |
| Latest release | v0.43.0 (2026-08-25) — apache-burr | langgraph v1.2.16 (2026-10-16) |
| License | Apache-2.0 | MIT |
| GitHub | apache/burr (formerly DAGWorks-Inc/burr) — ~2.6k stars, last commit 2026-08 | langchain-ai/langgraph — ~43k stars, last commit 2026-10 |
| Pricing | free (OSS) | OSS free (MIT); LangSmith Developer $0 (5k traces/mo, 1 seat); Plus $39/seat/mo (1 free small serverless deployment); deployments metered at $1/LSU by compute+memory; Enterprise custom |
| Funding | — | LangChain parent: $125M Series B at $1.25B valuation (2025-10); total not stated |
| Backend storage | — | pluggable (Postgres, SQLite, Redis) |
| Deployment | Self-hosted only — library + self-hosted tracking UI | Both (LangGraph OSS self-hosted; LangSmith Deployment cloud managed or self-hosted) |
| API surface | searched not found | SDK: Python, JS/TS |
| Embedding | — | multiple supported |
| Multi-tenancy | searched not found | — |
| MCP | searched not found | yes — Agent Server exposes deployed graphs as MCP tools at /mcp (streamable HTTP); MCP client via langchain[mcp] adapter |
| A2A | searched not found | supported — LangGraph A2A wrappers |
| OpenTelemetry | searched not found | first-class — LangSmith + OTel |
| Optimised for | production-grade state-machine LLM workflows with observability | durable graph-state checkpointing + Store System for cross-thread memory |
| Anti-fit | searched not found | not for non-LangGraph workflows |
Taxonomy
| Axis | Burr (DAGWorks) | LangGraph Persistence |
|---|---|---|
| storage | — | relational |
| retrieval | — | exact-match |
| persistence | — | cross-session |
| update | — | overwrite |
| unit | — | episode |
| governance | — | inspectable |
| conflict | — | last-write-wins |
Pros & cons
Burr (DAGWorks)
Pros: State-machine abstraction (vs DAG / ReAct) is genuinely more production-debuggable; companion to Hamilton (popular dataflow framework); strong founder pedigree.
Cons: Smaller stars / ecosystem than LangGraph / CrewAI; commercial sustainability of DAGWorks newer; state-machine framing has learning curve.
LangGraph Persistence
Pros: Persistent state for graph agents — checkpointing and resumability are first-class, not bolted on.
Cons: LangGraph-shaped; persistence model is general but most idiomatic when paired with LangChain memory.