Key takeaways
- Raised a $125M Series B at a $1.25B valuation (Oct 2025) and shipped LangChain 1.0 and LangGraph 1.0 — now ~$160M total disclosed funding
- #1 downloaded agent framework with 100M+ monthly downloads and 139k+ GitHub stars as of June 2026
- LangSmith expanded into a full agent platform at Interrupt 2026 — Engine (autonomous agent improvement), Sandboxes (GA), Fleet, and an LLM Gateway
FAQ
What is LangChain?
LangChain is a framework for building LLM-powered applications with a standard interface for models, embeddings, vector stores, and 1000+ integrations.
What is LangGraph?
LangGraph is LangChain's low-level orchestration framework for building stateful, multi-agent workflows with human-in-the-loop and durable execution.
How much does LangSmith cost?
LangSmith Developer is free (1 seat, 5k traces/month). Plus is $39/seat/month with 10k traces. Enterprise pricing is custom. Usage-based add-ons cover extra traces, deployment runs, Fleet runs, Engine compute, and sandboxes.
Is LangChain open source?
Yes, both LangChain and LangGraph are MIT-licensed open source. LangSmith (observability/deployment) is the commercial platform.
Who uses LangChain in production?
Klarna, LinkedIn, Uber, GitLab, Workday, Elastic, Rakuten, Replit, and thousands of other companies use LangChain products in production.
Executive Summary
LangChain is the #1 downloaded agent framework, with 100M+ monthly downloads and 139k+ GitHub stars as of June 2026. The company raised a $125M Series B at a $1.25B valuation in October 2025 and shipped LangChain 1.0 and LangGraph 1.0 alongside it. The ecosystem includes the LangChain framework for composable LLM applications, LangGraph for stateful agent orchestration, and LangSmith — now a full agent platform with observability, evaluation, deployment, sandboxes, and autonomous improvement. Trusted by Klarna, LinkedIn, Uber, and GitLab — and with 5 of the Fortune 10 as LangSmith customers — LangChain has become the default infrastructure layer for LLM application development.
| Attribute | Value |
|---|---|
| Company | LangChain Inc. |
| Founded | 2022 |
| Funding | ~$160M disclosed ($10M seed, $25M Series A, $125M Series B) |
| Valuation | $1.25B (Series B, Oct 2025) |
| Headquarters | San Francisco, CA |
Product Overview
LangChain provides the platform for building reliable AI agents, offering both high-level abstractions for rapid prototyping and low-level primitives for fine-grained control.
The ecosystem separates concerns across three products:
Key Capabilities
| Capability | Description |
|---|---|
| LangChain | Composable framework with 1000+ integrations for LLM apps (1.0 GA) |
| LangGraph | Low-level agent orchestration with state, memory, and control flow (1.0 GA) |
| LangSmith | Observability, evaluation, and deployment platform |
| LangSmith Engine | Clusters production failures, diagnoses root causes, proposes fixes (beta) |
| LangSmith Sandboxes | Isolated, hardware-virtualized code execution for agents (GA) |
| LangSmith Fleet | Prebuilt hosted agents (coding, GTM, research) with managed runtime |
| Deep Agents | Planning, memory, and sub-agents for complex long-running tasks; managed runtime available |
| LangGraph Platform | Scalable deployment infrastructure for agent workflows |
Product Surfaces / Editions
| Surface | Description | Availability |
|---|---|---|
| LangChain (Python) | Core framework for LLM applications | GA (1.0, Oct 2025) |
| LangChain (JS/TS) | TypeScript implementation | GA (1.0, Oct 2025) |
| LangGraph | Agent orchestration framework | GA (1.0, Oct 2025) |
| LangSmith | Observability and deployment | GA |
| LangSmith Engine | Autonomous agent improvement from production traces | Public beta |
| LangSmith Sandboxes | Secure agent code execution | GA |
| LangGraph Studio | Visual prototyping and debugging | GA |
Technical Architecture
LangChain provides a layered architecture from high-level chains to low-level graph-based workflows.
Languages: Python, TypeScript/JavaScript
LangGraph Architecture
LangGraph is inspired by Pregel and Apache Beam, providing a graph-based approach to agent orchestration:
from langgraph.graph import START, StateGraph
graph = StateGraph(State)
graph.add_node("node_a", node_a)
graph.add_node("node_b", node_b)
graph.add_edge(START, "node_a")
graph.add_edge("node_a", "node_b")
Key Technical Details
| Aspect | Detail |
|---|---|
| Deployment | Self-hosted, LangGraph Platform, cloud |
| Model(s) | OpenAI, Anthropic, Google, Azure, 40+ providers |
| Integrations | 1000+ integrations (vector stores, tools, retrievers) |
| Open Source | Yes (MIT License for LangChain/LangGraph) |
LangGraph Features
- Durable execution — Persists through failures, auto-resumes
- Human-in-the-loop — Inspect and modify agent state at any point
- Comprehensive memory — Short-term working memory + long-term persistent memory
- First-class streaming — Token-by-token streaming of agent reasoning
Strengths
- Market dominance — 100M+ monthly downloads, 139k+ GitHub stars (June 2026), #1 framework
- API stability — LangChain 1.0 and LangGraph 1.0 (Oct 2025) ended the pre-1.0 churn era with stable interfaces
- Enterprise adoption — Production use at Klarna, LinkedIn, Uber, GitLab, Workday
- Integration breadth — 1000+ integrations with every major LLM, vector store, and tool
- Developer experience — Excellent documentation, LangChain Academy courses, active community
- Full stack — From prototyping (LangChain) to orchestration (LangGraph) to production (LangSmith)
- Framework neutral observability — LangSmith works with any agent stack, not just LangChain
- Both languages — Full Python and TypeScript support
Cautions
- Complexity accumulation — Ecosystem has grown large; learning curve steeper than alternatives
- Abstraction overhead — Some developers find too many abstractions between code and LLM calls
- LangSmith dependency — Advanced observability requires paid platform
- Legacy migration debt — Pre-1.0 codebases still face migration work to the stabilized 1.0 APIs
- Performance concerns — Abstraction layers add overhead compared to direct API calls
- Vendor consolidation — Deep LangSmith integration may create switching costs
Pricing & Licensing
LangChain/LangGraph (Open Source)
| Tier | Price | Includes |
|---|---|---|
| Open Source | Free | Full framework (MIT License) |
LangSmith Platform
| Tier | Price | Includes |
|---|---|---|
| Developer | Free | 1 seat, 5k base traces/month, 1 Fleet agent (50 runs/month) |
| Plus | $39/seat/month | Unlimited seats, 10k base traces/month, 1 dev-sized deployment, 500 Fleet runs/month |
| Enterprise | Custom | SSO, hybrid/self-hosted, custom SLA, dedicated support |
Additional costs (as of June 2026):
- Base traces beyond quota: $2.50/1k; extended traces (long retention): $5/1k
- Deployment runs: $0.005/run; uptime: $0.0007/min (dev) to $0.0036/min (production)
- Fleet runs beyond quota: $0.05/run
- LangSmith Engine: $1.50 per LangChain Compute Unit (LCU)
- Sandboxes: CPU, memory, and storage billed per second
Hidden costs note: the platform now has five usage meters (traces, deployments, Fleet, Engine, sandboxes) — production bills are workload-dependent and harder to forecast than the $39 seat price suggests.
Competitive Positioning
Direct Competitors
| Competitor | Differentiation |
|---|---|
| CrewAI | CrewAI specializes in multi-agent teams; LangChain is broader LLM tooling |
| AutoGen | AutoGen focuses on research patterns; LangChain has production infrastructure |
| LlamaIndex | LlamaIndex excels at RAG; LangChain provides full agent development |
| Mastra | Mastra is TypeScript-native; LangChain supports both languages |
When to Choose LangChain/LangGraph Over Alternatives
- Choose LangChain/LangGraph when: You need maximum integrations, production observability, and the largest community
- Choose CrewAI when: You want simpler multi-agent team abstractions
- Choose AutoGen when: You need Microsoft ecosystem integration or research patterns
- Choose LlamaIndex when: Document understanding and RAG are your primary focus
Ideal Customer Profile
Best fit:
- Teams building production LLM applications requiring observability
- Organizations needing broad integration support (1000+ options)
- Developers wanting both Python and TypeScript
- Companies requiring enterprise features (SSO, compliance, support)
- Teams that value ecosystem and community over simplicity
Poor fit:
- Small projects where abstraction overhead isn't justified
- Teams wanting minimal dependencies and direct API calls
- Organizations avoiding vendor platform lock-in
- Projects where simplicity is more valuable than features
Viability Assessment
| Factor | Assessment |
|---|---|
| Financial Health | Strong — $1.25B valuation, ~$160M raised, 6k+ paying LangSmith customers |
| Market Position | Leader — #1 downloaded, highest mindshare, 5 of the Fortune 10 on LangSmith |
| Innovation Pace | Rapid — Engine, Sandboxes, Fleet, and SmithDB all shipped at Interrupt 2026 |
| Community/Ecosystem | Largest — 139k+ stars, 1M+ practitioners |
| Long-term Outlook | Strong — De facto standard for LLM development |
LangChain has achieved a dominant market position that creates network effects — more developers means more integrations means more developers. The risk is complexity growth making it harder for newcomers, but the ecosystem depth is also a significant moat.
Bottom Line
LangChain and LangGraph have earned their position as the default infrastructure for LLM application development. The combination of composable primitives, low-level orchestration control, and production-grade observability covers the full development lifecycle.
Recommended for: Teams building production LLM applications who value ecosystem depth, enterprise features, and the largest community. Especially strong for organizations needing observability and the flexibility to use LangChain, LangGraph, or just LangSmith with their own code.
Not recommended for: Small projects where simplicity matters more than features, or teams that want to minimize abstractions between their code and LLM APIs.
Outlook: With $125M of Series B capital and a $1.25B valuation, LangChain is converting framework dominance into a platform business: LangSmith Engine (autonomous agent improvement), Sandboxes, Fleet, and an LLM Gateway all shipped at Interrupt 2026. The key metric to watch is LangSmith monetization — 6k+ active customers and 5 of the Fortune 10 already, and that's the revenue engine.
Research by Ry Walker Research • methodology