LangGraph Pricing in 2026: Plans, Limits and Real Costs
LangGraph pricing lives inside LangSmith. Solo founders should start on Developer at $0 / seat per month with 5k base traces. Small teams deploying agents get the most from Plus at $39 / seat per month, which adds 10k base traces and one free small serverless deployment.
Verified 2026-08-21 · official pricing page · LangGraph in the directory
Quick pricing verdict
- Starting price
- $0 / seat per month on the Developer plan, then pay-as-you-go.
- Free plan
- Yes. Developer is $0 / seat per month, includes up to 5k base traces / mo, and is limited to 1 seat.
- Best plan for most people
- Plus at $39 / seat per month. It adds 10k base traces / mo, unlimited seats, 1 free small serverless deployment, Engine, and 25 LCU of Fleet usage / mo.
- Plan to avoid first
- Enterprise. Pricing is custom and built for advanced hosting, security, and support needs you probably do not need early on.
- Biggest hidden cost
- Pay-as-you-go charges beyond the included base traces, plus LCU and LSU metering for Deployment, Fleet, Engine, and Sandboxes once included allowances are used.
- Pricing checked
- 2026-08-21, against the official page
LangGraph is sold as part of the LangSmith platform under three plans: Developer ($0 / seat per month), Plus ($39 / seat per month), and Enterprise (custom pricing). Every plan charges pay-as-you-go on top of the seat fee.
All usage is metered in two units. An LCU (LangChain Compute Unit) covers work and compute at $1.50 / LCU. An LSU (LangChain Storage Unit) covers traces and storage at $1.00 / LSU. Specific products like Deployment, Fleet, Engine, and Sandboxes meter at their own per-resource rates.
Plans compared
| Plan | Price | Best for | Key limits |
|---|---|---|---|
| Developer | $0 / seat per month, then pay-as-you-go | Solo users getting started | Maximum of 1 seat; up to 5k base traces / mo; Community Forum support; no Deployment or Engine |
| Plus | $39 / seat per month, then pay-as-you-go | Teams building and deploying agents | Unlimited seats ($39 per seat/month); up to 10k base traces / mo; 1 free small serverless Deployment; 25 LCU of Fleet usage / mo included |
| Enterprise | Custom pricing, then pay-as-you-go | Teams with advanced hosting, security, and support needs | Custom seats and workspaces; self-hosted and hybrid deployment options; custom SSO, ABAC, and RBAC; Support SLA |
Source: LangGraph pricing page, checked 2026-08-21.
What each plan includes
Developer
- Includes
- 1 seat free, 5k base traces / mo, Community Forum support, 5 LCU of Fleet usage / mo free, 5 LCU and 1 LSU of Sandbox usage / mo free, limited to 10 Sandboxes, SSO via Google and GitHub
- Missing
- Deployment, Engine, LLM Gateway controls, Tuned Evaluators, Email and Slack support, architectural guidance, Support SLA
- Avoid if
- You need to deploy agents to production, run Engine, or share seats across a team
- Verdict
- Cheapest way to start. Pay-as-you-go kicks in for traces above 5k / mo and for metered Sandbox, Fleet, Deployment, and Engine usage.
Plus
- Includes
- Missing
- Custom SSO, ABAC, RBAC, hybrid or self-hosted deployment, Support SLA, access to deployed engineers
- Avoid if
- You only need a personal project with low trace volume, or you need self-hosted or hybrid deployment
- Verdict
- The most complete self-serve plan for teams. Additional deployments and usage beyond the included LCUs are billed separately.
Add-ons, overages, and surprise costs
| Cost item | Trigger | Published amount | Billing behaviour |
|---|---|---|---|
| Additional traces beyond the included base allowance | Usage above 5k (Developer) or 10k (Plus) base traces per month | 0.005 LSU / additional trace | Billed monthly in arrears as pay-as-you-go. |
| Additional Serverless or Dedicated deployments | Spinning up additional Serverless or Dedicated deployments on Plus or Enterprise | Metered by resource consumption: Runtime Compute 0.045 LCU / vCPU-hr, Runtime Memory 0.006 LCU / GiB-hr, Database Compute 0.177 LSU / vCPU-hr, Database Memory 0.025 LSU / GiB-hr | Charged based on resources consumed; billed per second. |
| Fleet usage beyond the included allowance | After 5 LCU (Developer) or 25 LCU (Plus) of Fleet usage per organization per month | Metered in LCUs at $1.50 / LCU | Additional usage is billed. Developer pauses new runs when the allowance is used until reset or upgrade to Plus. |
| Sandbox usage beyond the included allowance | After 5 LCU and 1 LSU of Sandbox usage per month | Compute 0.0384 LCU / vCPU-hr, Memory 0.0123 LCU / GiB-hr, Storage 0.000123 LSU / GiB-hr | All rates billed per second. |
| Engine runs | Each scheduled Engine run, which fires once every 6 hours | ~5–30 LCUs per run at $1.50 / LCU | Plus plans: billed as an additional charge on top of the monthly license fee. Enterprise: charged against committed LangChain credits. Includes underlying model usage. |
| Tuned Evaluators | Each successful Tuned Evaluator run | 0.01 LCU / Evaluation Run | Skipped and failed evaluations are not billed. Currently in public beta for Plus and Cloud Enterprise plans in the US. |
| Upgrading base traces to extended traces | Choosing 400-day retention instead of the default 14-day retention | Not published | An additional fee is charged to upgrade base traces to extended retention. |
| Third-party tools connected through Fleet | Authenticating with third-party tool accounts via remote MCP server | Not published | Any fees or usage charges are billed by the third-party provider. |
| LLM Gateway model usage | Pay-as-you-go model usage via Gateway Credits | Not published | Consumes prepaid Gateway Credits. BYO model API key is also supported. |
Source: Platform pricing docs, checked 2026-08-21.
Which plan should you choose?
- Solo founder prototyping a personal project
- Developer — $0 / seat per month covers 1 seat, 5k base traces / mo, Community Forum support, and 5 LCU of Fleet plus 5 LCU and 1 LSU of Sandbox per month at no charge.
- Small team building and deploying an agent
- Plus — $39 / seat per month adds unlimited seats, 10k base traces / mo, 1 free small serverless deployment, Engine, Tuned Evaluators, LLM Gateway controls, and 25 LCU of Fleet per organization each month.
- Team that needs self-hosting, hybrid hosting, custom SSO/ABAC/RBAC, or a Support SLA
- Enterprise — Only Enterprise offers self-hosted and hybrid deployment, custom SSO, ABAC, and RBAC, Support SLA, and access to deployed engineers.
Frequently asked questions
- Which plan is right for me?
- Developer suits personal projects with 1 free seat and 5k base traces / mo. Plus is for teams that want self-serve with unlimited seats, 10k base traces / mo, and 1 free small serverless deployment. Enterprise is for advanced administration, security, support, or deployment needs; contact sales.
- Do you offer a plan for startups?
- Yes. LangSmith for Startups offers discounted rates and credits for VC-backed startups, with up to $10,000 in credits.
- When will I be billed?
- For Developer or Plus, seats are billed monthly on the 1st or pro-rated if added mid-month with no credit for removed seats; traces are billed monthly in arrears. Enterprise plans are invoiced annually upfront.
- What is a seat?
- A seat is a distinct user inside your organization. The total number of users, including invited users, is used to determine the number of seats to bill.
- Can I upgrade from Developer to Plus or Enterprise?
- Yes. Developer plan users can upgrade to Plus on the Plans and Billing page. For Enterprise, contact the sales team.
- What is an LCU?
- An LCU (LangChain Compute Unit) is a normalized unit of work done on the LangSmith Platform used for pricing. Different services meter at different rates. 1 LCU = $1.50. Examples include Engine, Fleet, deployments, and sandboxes.
- What is an LSU?
- An LSU (LangChain Storage Unit) is a normalized unit of data stored or managed by LangSmith used for pricing. 1 LSU = $1.00. Examples include traces, deployments, and sandboxes.
- What is the difference between a base trace and an extended trace?
- Base traces have a 14-day retention period. Extended traces have a 400-day retention period and can be upgraded from base traces for an additional fee.
- Are Tuned Evaluations charged for every agent run?
- Tuned Evaluators are billed only when a Tuned Evaluator runs successfully. Skipped and failed evaluations are not billed. They are currently in public beta for Plus and Cloud Enterprise plans in the US.
- Does LangSmith Deployment include any free deployments?
- Plus plans include 1 free small serverless deployment. Additional serverless or dedicated deployments are charged on usage (resource time).
- Serverless vs. dedicated deployment?
- Serverless deployments scale to zero when idle and are intended for testing, background, or internal agents. Dedicated deployments are always on and intended for production workloads that require autoscaling, backups, and high availability.
- Are Fleet model costs included in Fleet pricing?
- Yes. Fleet pricing includes underlying model costs across Fast, Pro, and Max tiers.
- How is Fleet metered?
- Fleet usage is measured in LCUs. Developer includes 5 LCUs per organization each month; Plus includes 25 LCUs per organization each month. After the allowance, Developer pauses new runs until reset or upgrade; Plus bills additional usage.
- How is Engine priced and when does it run?
- Engine is metered in LCUs at $1.50 / LCU. An Engine run can consume ~5–30 LCUs. Engine is scheduled to run once every 6 hours. Pricing includes underlying model usage.
- How is Sandbox uptime calculated?
- Sandboxes are billed per second as ephemeral compute environments with configurable TTLs; inactive sandboxes shut down automatically. Sandboxes are available for SaaS Cloud deployments and not currently available for self-hosted deployments.
What we read
Every figure on this page was extracted from the vendor's own pricing page at the moment below. The captured text is one click away, so you can check our reading against what they publish today.
Source captured 2026-08-21 08:14 UTC · read the captured text
Pricing change log
What moved since we started tracking this page, newest first. A price that changes silently is the failure mode of every other pricing guide.
- 2026-08-21First published, extracted from https://www.langchain.com/pricing.