Pricing

The agent and the SDK are free and open source, with no Caveman account. Caveman Cloud's Free plan is $0 for one person and needs no card. You get every feature up to the monthly allowances. Add a card to go past them at standard rates, up to a $50 billing limit per product unless you set your own. You still pay your model provider for inference, on your own keys.

No. Without a card, the allowances are hard limits: a product pauses at its allowance until next month, as it would at a billing limit. Add a card when you want to keep going past them at standard rates.

$300 a month, and it includes $300 of usage every month, counted at the same standard rates as Free on events, agent minutes and routing. Past the included usage you pay standard rates, up to billing limits you set. Included usage does not roll over. If you move from Free during a month, the fee and the included usage are prorated together. There are no seats.

Team starts with a 7-day free trial. You add a card at checkout, and nothing is charged during the trial, including usage. When it ends, the $300 fee and its $300 of included usage start, prorated for the rest of that month. We email you before the trial ends. Move to Free during the trial and nothing is charged. Each organization gets one trial. If your Free plan already has a card, moving to Team skips the trial and bills from that day.

Enterprise is an annual contract, priced on the AI spend Caveman manages and never per seat. It runs hosted or in your own cloud.

We stop recording that product until the month resets, and we never charge past the limit. Gateway calls keep going to your provider unchanged; only their traces are not written. For agent runs, no new run starts and runs in flight finish. For routing, no decision is made and each request runs as asked. On Free, the allowance is the limit until you add a card; then each product starts with a $50 limit. On Team, agent runs start with a $150 limit past the included usage.

One recorded LLM call, span or eval score is one event. A call through the Caveman gateway counts one event per started 64 KB it sends, so a coding-agent request is several events.

On Free, Caveman may use your prompt text and recorded requests to improve its models. Requests are kept until you delete them or your retention window expires. Move to Team any time to turn sharing off. Enterprise never shares.

No. Token markup is 0%. Model calls run on your own provider keys, and your provider bills you for them directly. Caveman bills for events, agent minutes and routing decisions, plus the Team fee and any package.

Not necessarily. Caveman measured input tokens with Proxy + Skill; Elasticsearch Labs measured output tokens with elastic-caveman. They are separate studies and do not add together. Results vary by workload, and cache pricing, retries and recovery calls affect the final bill.

The agent and the SDK compress locally and keep the original bytes on your machine; the smaller request goes to your model provider as before. Cloud sees the traffic you route through it. A customer-owned install keeps that traffic in your own cloud account. Hosted Enterprise stores it on Caveman Cloud, redacted and encrypted, and an admin can switch storage off.

The agent sits around the coding agent you already use, like Claude Code or Codex. The SDK goes inside agents you build, as middleware for LangChain, the Vercel AI SDK and the OpenAI and Anthropic SDKs. Both are free, and they work on their own or together.

The skill, CLI, SDKs, middleware, extension and Browse are Apache-2.0. Releases before 3.0.0 keep the licence they shipped with. Caveman Code and the standalone cavemem package are MIT. Check each repository for the full terms.

Yes. Enterprise installs one Helm chart and the same signed images as hosted Cloud on your Kubernetes, on AWS or GCP. Your keys, KMS and storage stay in your account.