Pricing
Pay for Kelven.
Not unnecessary AI compute.
Kelven uses premium intelligence when it improves the result and more efficient models when it doesn’t.
Free
Experience the Kelven difference.
$0
Everything below comes with a free account, so your privacy choices and memory belong to you from the first message.
- The core Kelven experience
- Limited monthly intelligence usage
- Kelven Auto routing
- Memory that's yours, on your device or your account
- Every privacy mode — never paywalled
- Up to 2 projects
- See and pin which model answers
- MCP Gateway access (beta)
- Mobile and installable app access
- Recommended
Pro
For people who use AI every day.
$25/ month
Kelven automatically chooses when premium intelligence is worth using.
- Generous monthly intelligence allowance
- Premium model access
- Portable Kelven memory
- Multiple projects
- Advanced privacy controls and tokenization
- MCP Gateway access
- Multiple connected AI interfaces
- File analysis
- Web and research capabilities
- Routing transparency
- Usage and savings dashboard
- Bring your own model keys
Power
For people who push AI harder.
$60/ month
More high-compute intelligence when you actually need it.
- Everything in Pro
- Much larger intelligence allowance
- Priority access to expensive reasoning models
- Kelven Verify with a second model
- Multi-model workflows
- Deep Research
- Large files and larger context windows
- Higher-volume MCP usage
- Advanced routing controls
- Priority processing where appropriate
Free and paid — the difference
- Free (with an account): everything in the Free column above — your own memory, every privacy mode, projects, and the installable app. An account is always needed to use Kelven, so your memory and privacy settings belong to you from the first message.
- Pro and Power: a larger monthly allowance, premium models, second-opinion verification and deeper file work.
Privacy controls are part of every plan, including Free. Paid plans are billed monthly and can be changed or cancelled at any time from your billing settings.
Three different needs
One layer, priced the way each audience uses it.
People want a plan that simply works. Developers want to pay for exactly what they consume. Organisations want control over both.
For people
A plan buys you access to the whole intelligence layer — routing, memory, privacy and projects — without watching a meter.
For developers
You pay for what you actually consume through the API and MCP Gateway, and you can see what routing saved you.
For organisations
Control first: policy, provider rules and audit, on top of the consumption your teams use.
Developers, API and MCP
Pay for what you use. See what Kelven saves.
Developer access is usage-based: you pay for the intelligence your calls consume, and every response reports what it cost against the premium baseline it was compared with.
- One integration instead of one client library per provider
- Routing, privacy minimisation and tokenization applied to every call
- MCP Gateway access with scoped, revocable permissions
- Per-request cost and savings reporting
Per-request and per-token rates are published from the live billing configuration rather than estimated here, so the number you read is the number you are charged.
Enterprise
Enterprise starts at $4,500/month.
Includes up to 50 employees. Additional seats are $60 per employee per month. Billed annually. AI usage is additional.
Minimum 50 employees. Estimates shown up to 2,000; talk to us above that.
50 employees
$4,500/ month equivalent
$54,000/ year platform
- Platform fee, annual commitment, billed annually
- AI usage billed separately — not included, not unlimited
An estimate of the platform fee only. Inference rates are agreed in your contract, so nothing is guessed here.
What you get, and where it stands
Every capability carries one status, used the same way everywhere on this site: Available — working in the product today. Beta — working, still being hardened — expect rough edges. Planned — designed and on the roadmap, not built yet. By arrangement — delivered as part of a contract, not self-serve.
- Centralised AI routing with per-request transparency Available
- Privacy modes and provider eligibility rules Available
- Usage and efficiency visibility Available
- Billing and subscription management Available
- Bring your own provider keys Available
- Organisational context and memory controls Beta
- MCP Gateway with scoped, revocable read access Beta
- On-device (local) AI on supported hardware Beta
- Connected applications acting on your behalf Planned
- Priority onboarding and support By arrangement
- SAML/SSO and SCIM provisioning By arrangement
- Dedicated, VPC or on-premise deployment By arrangement
- Custom compliance review and data agreements By arrangement
- Custom integrations and connectors By arrangement
- Availability commitments and SLAs By arrangement