CrewAI pricing

About

CrewAI started in 2023 as an open-source Python framework for orchestrating role-playing, collaborative AI agents, built by João Moura while he was director of AI engineering at Clearbit. The MIT-licensed framework became one of the breakout projects of the agent era — over 50,000 GitHub stars by early 2026 — and by October 2024 the company said its OSS was executing more than 10 million agents per month.

In October 2024 CrewAI announced $18M in total funding — an inception round led by boldstart ventures and a Series A led by Insight Partners, with Craft Ventures, Earl Grey Capital, Blitzscaling Ventures, and angels including Andrew Ng and HubSpot CTO Dharmesh Shah — and launched CrewAI Enterprise, its managed cloud platform, after signing 150 beta enterprise customers in under six months. In October 2025 the enterprise product was rebranded CrewAI AMP (Agent Management Platform), spanning AMP Cloud (managed SaaS), AMP Factory (self-hosted on your own infra), and the OSS framework. The pricing page now claims usage by 63% of the Fortune 500.

For the most current information, visit CrewAI.

Pricing summary : How CrewAI’s pricing model works

CrewAI prices its AMP platform on workflow executions across two tiers. Basic is free: the Studio visual editor with AI copilot, GitHub integration, unlimited deployments, tracing, guardrails, and a usage dashboard — but a hard cap of 50 workflow executions per month with no overage path. Enterprise is custom-quoted: the included execution allotment is “sized to workflow,” overage beyond it is “flexible,” and the package layers on private infrastructure options (dedicated VPC, NAT, SAM-certified, FedRAMP High), SSO and RBAC, dedicated support, and an unusual 50 hours of development per month from CrewAI’s own team.

The platform fee covers orchestration, deployment, and monitoring — not model inference. You connect your own LLM provider keys, so OpenAI, Anthropic, or other token bills arrive separately. And the framework itself is MIT-licensed: teams that want pure code-first orchestration can self-host for free with no execution cap.

What makes this different: CrewAI ran the transparency arc in reverse. It launched public pricing in October 2025 (a free tier, a low-cost self-serve Professional plan, and Enterprise — all with a published per-execution overage rate), then deleted the mid-tier in spring 2026, then deleted the published overage rate itself — ending up with a public page whose only number is “Free.” The withdrawn figures are documented in Pricing evolution below.

Pricing by product

Sales motions across products: self-serve signup for Basic, open-source self-host for the framework, and sales-led for Enterprise (“Request trial” / “Meet with us”). LLM token costs are always pass-through — you bring your own provider keys on every tier.

Hidden costs : What CrewAI users actually pay

The platform price is only half the bill. CrewAI AMP meters workflow executions, but every execution also burns LLM tokens billed directly by your model provider — a multi-agent crew can make dozens of model calls per run, so token spend routinely exceeds the platform fee. On Enterprise, the real negotiation is the size of the included execution allotment and the overage rate, neither of which is published anymore (the last public rate, $0.50/execution, disappeared from the page between May and June 2026).

Budget watch-outs: the free tier’s 50-execution cap is a demo allowance, not a production runway — one daily scheduled workflow exhausts it in under two months of runs; and the bundled 50 development hours/month signal that Enterprise contracts are services-heavy engagements, not self-serve SaaS.

Want to estimate your own CrewAI bill? Use the CrewAI pricing calculator to model your costs based on usage patterns.

Pricing evolution : CrewAI pricing history and changes

Cadence

Tracked range: 2024–present, via the launch press, the 2025-10-03, 2026-01-19, 2026-02-25, 2026-03-05, 2026-04-07, 2026-05-03 and 2026-06-03 Wayback snapshots, and a live 2026-06-10 capture.

Notable changes

• 2024-10 — CrewAI Enterprise launches alongside the $18M announcement. Entirely sales-led; no public pricing page exists.
• 2025-10 — Rebrand to CrewAI AMP (“the first Agent Management Platform”) and first public pricing: Basic free (50 executions/mo), Professional $25/mo (100 executions, $0.50 per additional), Enterprise custom (up to 30,000 included executions, $0.50/execution overage).
• 2026-02 — Professional reframed around +1 seat and community-forum support; Basic loses its overage path (hard 50 cap). Enterprise messaging pivots to compliance: SOC2, SSO, PII detection/masking, Forward Deployed Engineers.
• 2026-04/05 — Professional removed (between the 2026-04-07 and 2026-05-03 snapshots). The page returns to Free + Enterprise.
• 2026-05/06 — The published $0.50/execution overage disappears (between the 2026-05-03 and 2026-06-03 snapshots): Enterprise executions become “sized to workflow” with “flexible overage.” Confirmed live on 2026-06-10.

What’s unique : CrewAI’s distinctive pricing mechanics

1. Workflow executions as the value meter. CrewAI bills the platform on completed workflow runs, not seats, tokens, or API calls. One execution can hide dozens of agent steps and LLM calls, so the meter sits close to “jobs done” — easy to count, easy to explain to a CFO — while the volatile token cost underneath is passed through to your own provider keys.

2. A full transparency reversal. Most corpus companies move toward published pricing as they mature. CrewAI went the other way: sales-led (2024) → fully published three-tier pricing with a $0.50 unit rate (Oct 2025) → mid-tier deleted (spring 2026) → unit rate deleted (mid-2026). The public page now contains exactly one price: Free.

3. Services bundled into the platform SKU. Enterprise includes 50 hours of development per month, up to 100 hours of onboarding, on-site training, and Forward Deployed Engineers. That is consulting capacity sold inside a SaaS subscription — a tacit admission that production multi-agent systems still need expert builders, and a way to anchor a much higher contract value than software alone.

Strengths & weaknesses

Billing UX : CrewAI billing controls and transparency

• Billing controls — Basic is self-serve signup with a hard execution cap, so there is no surprise-bill risk on free: workflows simply stop at 50 executions. Enterprise is contract-based (“Request trial” / “Meet with us”), with allotments and overage negotiated per deal.
• Usage visibility — The platform ships a usage dashboard with execution counts, token counts, performance metrics, hallucination scores, and deployment history on both tiers, plus tracing and OpenTelemetry export. Forward cost visibility is good for executions, weaker for the pass-through token spend that lands on your provider invoices.
• Payment options — Free tier requires no card. Enterprise is invoiced, with deployment on CrewAI cloud or customer infrastructure (dedicated VPC, NAT, SAM-certified, FedRAMP High) and support via Slack/Teams, on-site, and dedicated contacts.

Strategic wins : Why CrewAI’s pricing decisions worked

1. The OSS framework as a Fortune 500 funnel

The MIT-licensed framework (50K+ GitHub stars, 10M+ agent executions/month by late 2024) seeded CrewAI inside enterprises before a sales team existed — the company signed 150 beta enterprise customers in under six months and now claims 63% of the Fortune 500. The free framework removes the evaluation barrier; AMP monetizes deployment, governance, and support. See how AI companies structure pricing.

2. Picking executions over seats or tokens

Metering workflow executions ties the platform fee to completed automation rather than headcount (agents replace seats, so seat pricing would shrink with success) or raw tokens (which buyers can’t predict). It also cleanly separates CrewAI’s orchestration value from pass-through model costs. See choosing the right usage metric.

3. Bundling development hours into Enterprise

Production multi-agent systems are still hard to build. Folding 50 hours/month of CrewAI’s own engineers into the subscription converts a risky software bet into a guided rollout — increasing win rates, deal sizes, and switching costs in one move. Related: outcome-based pricing trends.

Areas to improve : Gaps in CrewAI’s pricing approach

1. The transparency reversal erodes trust

Publishing $0.50/execution in October 2025 set an anchor; deleting first the $25 tier and then the rate itself leaves prospects with no way to budget without a sales call. Buyers who saw the old page will assume the unpublished price is higher. See bill shock and cost unpredictability.

2. The missing middle

Between a free tier capped at 50 executions and a services-heavy custom Enterprise contract, there is now nothing. Startups and mid-market teams that outgrow Basic in weeks have no self-serve paid path — exactly the segment the deleted $25 Professional tier served, and a gap competitors with credit-card pricing can exploit.

3. Total cost of ownership is two invoices

Because LLM tokens pass through on customer keys, a CrewAI quote never shows the full cost of running agents. Surfacing blended cost-per-execution estimates (the dashboard already counts tokens) would make the economics legible and defuse procurement surprises.

Key takeaways

1. Open source can carry enterprise GTM. An MIT framework with 50K+ stars put CrewAI inside the Fortune 500 before pricing existed; the paid platform monetizes governance and deployment, not the core capability.
2. Meter the work, pass through the inputs. Workflow executions price the outcome-shaped unit; volatile token costs stay on the customer’s own provider keys.
3. Transparency can go backwards. CrewAI published a unit rate, then withdrew the mid-tier and the rate within nine months — pricing pages are positioning documents, and retreats are as informative as launches.
4. Killing the bridge tier is a bet on enterprise. Removing the $25 Professional plan abandons self-serve revenue to focus the funnel on quoted contracts — and leaves a visible gap for rivals.
5. Services inside the SKU signal category immaturity. Bundled monthly development hours show that agent platforms still sell expertise alongside software in 2026.

UBP implications

1. Execution-based pricing is emerging as the agent-platform default. A completed workflow run is countable, demoable, and closer to value than tokens or seats — but it only works with clear definitions of what counts as one execution.
2. Hard caps are an alternative to overage on free tiers. CrewAI’s free plan simply stops at 50 executions — no card, no surprise bill — trading conversion friction for trust.
3. If you publish a unit rate, plan to live with it. Anchors persist in archives and buyer memory; withdrawing published prices reads as a price increase even if quotes haven’t changed. See usage-based pricing strategy.

Sources

• CrewAI pricing page — live capture (accessed 2026-06-10)
• CrewAI launches multi-agentic platform — $18M funding announcement (accessed 2026-06-10)
• Insight Partners on the CrewAI Series A (accessed 2026-06-10)
• CrewAI AMP — the Agent Management Platform (launch post, 2025-10-02) (accessed 2026-06-10)
• crewAI on GitHub (MIT, 50K+ stars) (accessed 2026-06-10)
• CrewAI AMP docs (accessed 2026-06-10)
• Wayback Machine snapshots: crewai.com/pricing — 2025-10-03, 2026-01-19, 2026-02-25, 2026-03-05, 2026-04-07, 2026-05-03, 2026-06-03 (accessed 2026-06-10)

Bottom line

CrewAI pairs one of the most popular open-source agent frameworks (MIT, 50K+ GitHub stars) with CrewAI AMP, an enterprise Agent Management Platform metered on workflow executions. Today’s pricing is stark: a free Basic tier hard-capped at 50 executions/month, and a custom-quoted Enterprise plan with executions “sized to workflow,” flexible overage, private-infrastructure options up to FedRAMP High, and 50 bundled development hours a month — with LLM token costs always passing through on your own keys. The $25 Professional tier and the published $0.50/execution rate that launched with the October 2025 AMP rebrand were both withdrawn by mid-2026, a rare full transparency reversal. Browse the pricing blueprint for more fully-researched company profiles.

Want to compare CrewAI against other agent-platform companies? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Published $0.50 overage rate removed — 'flexible overage'

Jun 2026

Live capture confirms two tiers: Basic free (50 executions/mo, hard cap) and Enterprise custom with included executions 'sized to workflow' and unpublished 'flexible overage' — the $0.50/execution rate vanished between the 2026-05-03 and 2026-06-03 snapshots. Enterprise bundles 50 hours of development/month, FedRAMP High options, SSO and RBAC.

Professional $25 tier removed — back to Free + Enterprise

May 2026

The $25/month Professional plan disappears (between the 2026-04-07 and 2026-05-03 Wayback snapshots), leaving only free Basic and custom Enterprise. The $0.50/execution overage rate is still published on the Enterprise column at this point.

Professional repackaged; Enterprise leads with compliance

Feb 2026

Professional ($25/mo) reframed as 'everything in Basic plus 1 additional seat' with 100 executions and community-forum support; Basic loses its overage path (hard 50-execution cap). Enterprise messaging now leads with SaaS or self-hosted K8s/VPC deployment, SOC2, SSO, PII detection and Forward Deployed Engineers.

AMP rebrand + first public pricing: Free / $25 Professional / Enterprise

Oct 2025

CrewAI Enterprise becomes CrewAI AMP (Agent Management Platform) and a public pricing page appears: Basic free (50 executions/mo), Professional $25/mo (100 included executions, $0.50 per additional), and custom Enterprise with up to 30,000 included executions at a published $0.50/execution overage.

CrewAI Enterprise launches — sales-led, no public prices

Oct 2024

Alongside an $18M raise (inception round led by boldstart, Series A led by Insight Partners), CrewAI launches its enterprise cloud subscription, CrewAI Enterprise, after signing 150 beta enterprise customers in under six months. Pricing is entirely sales-led with no public pricing page.