Simple, transparent pricing.
Self-host for free with all features included. Upgrade to managed cloud or enterprise on-prem when you need SLAs, compliance, and dedicated support.
Free
Free
All core features. No usage caps. Run it yourself.
- Full Gateway (all routing strategies, guardrails, semantic cache)
- Full Workspace (team chat, RAG, agents, voice, evals)
- Deploy module (Kubernetes + dstack, model library, fine-tuning)
- Unlimited API keys and teams
- Community support via GitHub Discussions
Team
Contact us
Managed hosting, priority support, and usage quotas that scale with your team.
- Everything in Free
- Managed Postgres, vector store, and object storage
- Per-seat and per-token usage quotas with auto-alerts
- SSO (SAML / OIDC) and SCIM directory sync
- 99.9% uptime SLA with dedicated support channel
- Audit log export to your SIEM or S3 bucket
Enterprise
Contact us
License file activation, air-gapped operation, and enterprise-grade compliance.
- Everything in Team
- License file activation — no internet required after setup
- Fully air-gapped deployment option
- SSO, SCIM, and role-based access control
- Dedicated customer success and SLA with response-time guarantees
- Custom data-processing agreements (DPA/BAA available)
No hidden fees. No per-token markup.
ManyLayers does not mark up provider API costs. You pay your providers directly using your own API keys. The Free tier is genuinely unlimited — your only cost is the cloud or hardware you run it on. Enterprise licenses are priced per deployment (not per seat or per token).
Frequently asked questions.
What is the difference between SaaS, hybrid, and self-hosted deployments?
SaaS (managed cloud) means ManyLayers operates the full infrastructure stack — control plane, data plane, databases, and vector stores — on your behalf. Hybrid keeps the control plane managed while your data plane runs inside your own VPC, so prompts and completions never leave your network. Self-hosted is a full on-premises deployment: you install ManyLayers on your own Kubernetes cluster or bare-metal servers, activate it with a license file, and optionally operate it completely air-gapped with no outbound internet traffic after the initial setup.
How do per-team budgets work?
ManyLayers lets you define spend budgets at the team, project, or API-key level. Each budget has a configurable spend ceiling (in dollars or tokens), a reset cadence (daily, weekly, or monthly), and an alert threshold. When a team approaches its ceiling, ManyLayers sends an alert and optionally soft-blocks further requests until the period resets or an admin raises the limit. All budget events are captured in the immutable audit log for compliance review.
Which model providers and models are supported?
ManyLayers supports 40+ providers out of the box, including OpenAI, Anthropic, Amazon Bedrock, Google Vertex AI, Azure OpenAI, Gemini, Cohere, Mistral AI, Groq, DeepSeek, Together AI, Ollama, and all major open-weight model families (Llama, Qwen, Phi, Gemma, and more). The provider list is updated continuously; any OpenAI-compatible endpoint can also be added as a custom provider.
Can we run ManyLayers fully air-gapped?
Yes. The self-hosted Enterprise tier supports fully air-gapped operation. After initial installation and license file activation you can disable all outbound internet connectivity. Model weights are loaded from your internal artifact registry, telemetry is disabled by default, and all audit logs are written to local storage or your own SIEM. Air-gapped mode is certified for deployment in classified and high-security environments.
How does licensing work for self-hosted deployments?
The self-hosted Free tier is available to any organization and includes all core features with no usage caps. Enterprise self-hosted deployments are licensed per deployment instance, not per seat or per token, and are activated via a cryptographic license file. Licenses are scoped to your organization and may include options for unlimited replicas within your infrastructure. Contact us for pricing and to obtain a license file for enterprise on-prem use.
How do we get support?
Self-hosted Free users can access community support via GitHub Discussions and our public documentation at docs.manylayers.io. Team (SaaS) customers receive priority support through a dedicated Slack or Teams channel with a defined response SLA. Enterprise customers are assigned a dedicated customer success manager, have access to a private support portal, and receive a contractual response-time guarantee. For urgent issues outside business hours, Enterprise contracts include an on-call escalation path.