Fleet 1.13:Teams are now shipping 5x more PRs with autonomous pipelines.See what's new →
FleetFleet
Comparison

Fleet vs. Grok Build: Self-Hosted Fleet Orchestration vs. xAI's Cloud Agent

Grok Build is xAI's cloud-hosted coding agent backed by the Grok model family. Fleet is a self-hosted orchestration layer for multi-role teams of Claude Code agents, with per-role Claude model selection, governance, and no dependency on xAI's infrastructure.

Grok Build is xAI's entry into the autonomous coding agent space, launched in early beta in May 2026 and available to SuperGrok and X Premium+ subscribers. It uses Grok models to plan and implement software tasks from natural language descriptions and targets developers looking for an alternative to GitHub Copilot or Claude-based agents.

Fleet takes a different shape. It coordinates a team of specialized Claude Code agents — developer, reviewer, release manager, PM — each assigned the most appropriate Claude model tier for its role: Opus for judgment-heavy work like architecture and review, Sonnet for procedural development. Fleet handles the coordination and governance across that team; it does not run Grok models.

Choose Fleet if

Teams that want self-hosted, multi-role orchestration of Claude Code agents — choosing the right Claude model tier per role — with governance and no dependency on xAI's cloud.

Choose Grok Build if

Developers who want a cloud-hosted autonomous coding agent backed by xAI's Grok models with minimal setup.

Fleet vs. Grok Build: side by side

FeatureFleetGrok Build
DeploymentSelf-hosted binary on your infrastructurexAI cloud service
Model selectionPer-agent Claude model (Opus/Sonnet) chosen by roleGrok model family (xAI)
Multi-agent supportFull role-based roster with handoffsSingle agent per task
Agent handoffsFabric event bus with reactive chainNot supported
Data residencySelf-hosted; your source code goes only to your model backend (Bedrock/Vertex in your own cloud) + GitHubProcessed on xAI infrastructure
GovernanceBudget caps, risk scoring, quarantine, audit logNot exposed to operators
Watcher daemonReacts to GitHub labels automaticallyManual task assignment
AvailabilityGenerally available at fleetctl.aiEarly beta (launched May 2026) via SuperGrok / X Premium+

Where Fleet is the better fit

  • Per-role Claude model selection — assign Opus to judgment-heavy roles like architecture and review, Sonnet to procedural development
  • Self-hosted with full audit trail and no code processed by xAI's infrastructure
  • Multi-role pipeline automation that Grok Build's single-agent model does not support
  • Production-ready with versioned releases and active maintenance

Where Grok Build is the better fit

  • Zero setup — connect GitHub and start; no binary or config files required
  • Backed by xAI's significant model investment, with potential improvements as Grok models mature
  • Simple pricing model if you are already in the xAI/X ecosystem
  • Good option for developers who want to evaluate Grok's coding capabilities without building an orchestration setup

Pricing

Grok Build is bundled into xAI subscription tiers — access comes with SuperGrok and X Premium+ rather than standalone per-seat pricing. Fleet Team is $49/agent slot/month with a free tier. For production multi-agent pipelines, Fleet's cost is predictable on a per-slot basis.

Do they compete, or coexist?

The tools do not directly interoperate. Fleet runs Claude Code agents on your own infrastructure, while Grok Build runs Grok-backed agents in xAI's cloud. A team could use Grok Build for one-off cloud tasks and Fleet for its governed, multi-role pipeline, but Fleet does not run Grok models.

Frequently asked questions

Is Grok Build available for general use?

Grok Build launched in early beta in May 2026, available to SuperGrok and X Premium+ subscribers. Fleet is generally available at fleetctl.ai.

Can Fleet agents use Grok models?

No. Fleet's agent runner is Claude Code, so Fleet agents run on Claude models. Model selection in Fleet means choosing the Claude tier — Opus or Sonnet — per agent role, not swapping in Grok or other vendors. If you want to use Grok, run Grok Build directly.

More comparisons

Fleet vs. Google Jules →Fleet vs. Devin →Fleet vs. Codex CLI →

Run your first agent fleet

One binary. Five minutes. See every agent, coordinate every handoff, and keep a full audit trail of what your fleet did.

See how it worksInstall Fleet