---
title: "Caveman pricing"
description: "Caveman Code is free on your own keys and logins. Pro is $20 a month with $20 of credits, which buy up to $60 of usage on GLM-5.3. The open-source Caveman CLI is free: npx @caveman-ai/cli. Caveman Cloud is for teams building their own agents. Free for one person."
canonical: https://caveman.so/pricing
last-updated: 2026-11-16
---

# Caveman pricing

Caveman Code is free on your own keys and logins. Pro is $20 a month with $20 of credits, which buy up to $60 of usage on GLM-5.3. The open-source Caveman CLI is free: `npx @caveman-ai/cli`. Caveman Cloud is for teams building their own agents. Free for one person. Your whole team for $300 a month, $300 of usage included.

Caveman Code, our coding agent, is free on your own keys and logins; plans start at $20 a month with $20 of credits, and a credit goes up to three times as far on some models. The open-source Caveman CLI is free: install it with npx @caveman-ai/cli. Caveman Cloud is for teams building their own agents. Its Free plan has every feature for one person up to the monthly allowances, with no card. Team is $300 a month with $300 of usage included, then standard rates, and starts with a 7-day free trial.

## Caveman Code plans

Every plan is credits. On GLM-5.3 a credit goes three times as far, and twice as far on DeepSeek Flash and GLM-5.3 Flash. Every other model counts at list price.

### Free: $0 a month

For the keys and logins you already pay for.

- Your own keys and logins
- Top up from $5 for our models and Cloud computers
- 1 Cloud computer kept
- 16 GB of memory awake at once
- 4 CPU / 16 GB machines at most

[Start free](https://app.caveman.so/signup?next=%2Fcode%2F)

### Pro: $20 a month

For everyday coding. Or $192 billed yearly, $16 a month.

- $20 of credits every month
- Up to $60 on GLM-5.3 (3×)
- 10 Cloud computers kept
- 160 GB of memory awake at once
- 4 CPU / 16 GB machines at most
- 1 free month of Pro to give away

[Get Pro](https://app.caveman.so/signup?next=%2Fcode%2F)

### Pro $60: $60 a month

For coding most of the day. Or $576 billed yearly, $48 a month.

- $60 of credits every month
- Up to $180 on GLM-5.3 (3×)
- 20 Cloud computers kept
- 320 GB of memory awake at once
- 8 CPU / 32 GB machines at most
- 1 free month of Pro to give away

[Get Pro $60](https://app.caveman.so/signup?next=%2Fcode%2F)

### Max $100: $100 a month (recommended)

For several agents at once. Or $960 billed yearly, $80 a month.

- $105 of credits every month
- Up to $315 on GLM-5.3 (3×)
- 30 Cloud computers kept
- 384 GB of memory awake at once
- 16 CPU / 64 GB machines at most
- 2 free months of Pro to give away

[Get Max $100](https://app.caveman.so/signup?next=%2Fcode%2F)

### Max $200: $200 a month

For agents that run all day. Or $1,920 billed yearly, $160 a month.

- $220 of credits every month
- Up to $660 on GLM-5.3 (3×)
- 40 Cloud computers kept
- 416 GB of memory awake at once
- 16 CPU / 64 GB machines at most
- 3 free months of Pro to give away

[Get Max $200](https://app.caveman.so/signup?next=%2Fcode%2F)

Credits are dollars, spent at each model's list price, and reset every month. One week, or any 5 hours, can spend at most 35% of a month's credits, so one long day can't use the month up. Need more? Top-ups are $5 to $500 at a time, are spent at list price on every model, and last 365 days.

### What a plan's credits buy

| Models | A credit goes |
| --- | --- |
| GLM-5.3 | 3× as far |
| DeepSeek Flash, GLM-5.3 Flash | 2× as far |
| Claude Opus 5.5, Claude Sonnet 5.5, GPT-6.1 Sol, GPT-6 Luna, Claude Haiku 5.5, Kimi K3, DeepSeek V4 Pro, MiniMax M3 | List price |

## Caveman Cloud Free: $0 / month

Includes caveman /cloud. Every feature for one person. For one developer trying everything Caveman does.

Free every month:

### caveman /cloud

Hosted at app.caveman.so.

- Every Cloud feature: Observability, evals, monitors, detection, the inbox, Autopilot, Ask and agent runs · the semantic cache is on Team
- 100,000 events a month: LLM observability · recorded LLM calls, spans and eval scores
- 1 event per started 64 KB sent: A gateway call counts one event per started 64 KB it sends, so a coding-agent request is several events
- 600 agent minutes a month: Agent runs · runner time Caveman spends on your agents
- 25,000 routing decisions a month: Routing · one decision answered is one unit
- $10 of Caveman AI a month: For Caveman's own agents, like Ask and Autopilot, on the Luna model
- Standard rates past the allowances: Add a card to go past them · at most $50 a month per product unless you raise the limit
- One person: Move to Team to invite your team

[Start free](https://app.caveman.so/signup?next=%2Fcode%2F)

## Caveman Cloud Team: $300 / month

Includes caveman /cloud. Includes $300 of usage every month. For teams that want agent costs cut, with proof.

Everything in Free, plus:

### caveman /cloud

Hosted at app.caveman.so.

- $300 of usage every month: Events, agent minutes and routing, at standard rates. It does not roll over
- Then standard rates: The same rates as Free, past the included usage · events from $0.00035
- Your whole team: Invite members, the Team view and team settings · no seats
- $100 of Caveman AI a month: For Caveman's own agents, on Luna and Sol
- Semantic cache: A request that asks the same thing in different words is answered from cache
- Guardrails, budgets and an audit log: Project access control · Audit log export · Guardrails, budgets, access rules and key limits · Audit log · included, no package needed
- Content sharing off: Moving to Team stops Caveman using your content to improve its models
- SSO with the Scale package: $750 a month · SSO (OIDC and SAML) · Require SSO · IdP group mappings

[Try Team free for 7 days](https://app.caveman.so/signup?next=%2Fcode%2F)

## Caveman Cloud Enterprise: Custom / year

Includes caveman /cloud in your VPC. Annual contract. For orgs that keep data in their own cloud.

Everything in Team, plus:

### caveman /cloud

In your VPC, one Helm chart on AWS or GCP.

- Runs in your VPC: One Helm chart on your Kubernetes, AWS or GCP
- Your KMS and storage: Keys, traces and artifacts stay in your cloud account
- Signed releases: The same signed images as hosted Cloud, verifiable before you install
- Annual contract: Priced on the AI spend Caveman manages, never per seat

[Talk to us](https://cal.com/caveman/chat)

## Usage past the allowances

### LLM observability: 100,000 events free every month

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

| Monthly events | Per event |
| --- | --- |
| 0–100K | Free |
| 100K–150K | $0.00035 |
| 150K–250K | $0.000175 |
| 250K–500K | $0.0001 |
| 500K–1M | $0.00008 |
| 1M–10M | $0.00007 |
| 10M–50M | $0.000065 |
| 50M+ | $0.00006 |

Default billing limit: $50 on Free with a card; on Team, none until you set one. Change it any time.

### Agent runs: 600 minutes free every month

Minutes Caveman's runners spend on your agents, each run rounded up to a whole minute. Model calls stay on your own keys.

| Monthly minutes | Per minute |
| --- | --- |
| 0–600 | Free |
| 600–10K | $0.02 |
| 10K–100K | $0.015 |
| 100K+ | $0.012 |

Default billing limit: $50 on Free with a card; on Team, $150 past the included usage. Change it any time.

### Routing: 25,000 decisions free every month

One routing decision answered is one unit. The request itself goes to your provider, on your key.

| Monthly decisions | Per decision |
| --- | --- |
| 0–25K | Free |
| 25K–2M | $0.001 |
| 2M–10M | $0.0006 |
| 10M–50M | $0.0004 |
| 50M+ | $0.0003 |

Default billing limit: $50 on Free with a card; on Team, none until you set one. Change it any time.

At a billing limit we stop recording. We never block your traffic.

## Compare Cloud plans

| | Free | Team | Enterprise |
| --- | --- | --- | --- |
| **caveman /cli** | | | |
| Output compression | Yes | Yes | Yes |
| Content-aware compression | Yes | Yes | Yes |
| Model routing | soon | soon | soon |
| Waste fixes | Yes | Yes | Yes |
| Reusable scripts | soon | soon | soon |
| **caveman /middleware** | | | |
| Content-aware compression | Yes | Yes | Yes |
| Byte-exact recovery | Yes | Yes | Yes |
| Tool-schema compression | Yes | Yes | Yes |
| **caveman /cloud** | | | |
| Observability | Yes | Yes | Yes |
| Harness optimization | dev | dev | dev |
| Model fit per task | Yes | Yes | Yes |
| Evals | Yes | Yes | Yes |
| Agent-ready | Yes | Yes | Yes |
| Semantic cache | No | Yes | Yes |
| **Usage and billing** | | | |
| LLM observability: events | From $0.00035 | From $0.00035 | By contract |
| Agent runs: minutes | From $0.02 | From $0.02 | By contract |
| Routing: decisions | From $0.001 | From $0.001 | By contract |
| Included usage | No | $300 a month | By contract |
| Default billing limit | $50 per product with a card | $150 on agent runs | By contract |
| Caveman AI | $10 a month · Luna | $100 a month · Luna and Sol | $100 a month · Luna and Sol |
| Members | One person | Your whole team | Your whole team |
| Content used to improve Caveman's models | Yes | No | No, locked off |
| Token markup | 0% | 0% | 0% |
| **Deployment and control** | | | |
| Hosted at app.caveman.so | Yes | Yes | Yes |
| Runs in your VPC | No | No | Yes |
| Scoped keys | Yes | Yes | Yes |
| Data retention you set | Yes | Yes | Yes |
| Project access control | No | Yes | Yes |
| Audit log export | No | Yes | Yes |
| Guardrails, budgets, access rules and key limits | No | Yes | Yes |
| Audit log | No | Yes | Yes |
| SSO (OIDC and SAML) | No | Scale · $750/mo | Yes |
| SCIM | No | No | Yes |

## Published results

| Result | Source |
| --- | --- |
| 33.2% fewer input tokens | [Proxy + Skill study](https://github.com/JuliusBrussee/caveman/blob/main/docs/WRAP-BENCHMARK.md) |
| 63.6% fewer output tokens | [Elasticsearch Labs](https://www.elastic.co/search-labs/blog/elastic-caveman-ai-token-reduction) |

Proxy + Skill benchmark: six Claude Code tasks, three paired runs each. Both groups passed 18/18 exact-answer checks. Input-token reduction, not a promised reduction in your bill.

Elasticsearch Labs measured elastic-caveman across eight live MCP scenarios: 1,284 normal output tokens versus 467 Caveman output tokens. This is a separate study and workload from the Proxy + Skill input benchmark.

## Programs

### Prepaid and discounts

Prepaid usage, and startup or non-profit discounts, are set up by hand on Team and Enterprise. Tell us about your team and your traffic.

[Book a call](https://cal.com/caveman/chat)

### Open source

The agent and the SDK are Apache-2.0. Use them at work, read every line, fork them. Model-provider charges still apply, on your own keys.

[Browse the code](https://github.com/JuliusBrussee/caveman)

### Honest numbers

Every figure on this page names its source and its workload, including where Caveman saves nothing. Measured, inferred and verified savings stay separate labels.

[Read how we measure](https://caveman.so/news/does-the-caveman-skill-save-money)

## Questions

### How do Caveman Code plans work?

A plan is a monthly amount of credits. Credits are dollars, spent at each model's list price: Pro is $20 a month with $20 of credits, Pro $60 is $60 a month with $60 of credits, Max $100 is $100 a month with $105 of credits and Max $200 is $200 a month with $220 of credits. Paying yearly costs 20% less for the same credits every month. Caveman Code is also free on your own keys and logins.

### Which models make my credits go further?

On GLM-5.3 a credit goes three times as far, and twice as far on DeepSeek Flash and GLM-5.3 Flash. Every other model counts at list price. So Pro's $20 of credits buys up to $60 of GLM-5.3 in a month.

### Is there a weekly limit on a Caveman Code plan?

One week, or any 5 hours, can spend at most 35% of a month's credits, so one long day can't use the month up. Past a limit, top-up credits pay if you have any, until the plan resets.

### What is a top-up?

Extra credits you buy when you need them, on any plan or on Free. Top-ups are $5 to $500 at a time, are spent at list price on every model, and last 365 days. They have no weekly limit. Only the credits that come with a plan go further on some models.

### How are Cloud computers charged?

A Cloud computer draws on the same credits while it is awake: $0.12 an hour for 2 CPU / 8 GB, $0.20 an hour for 4 CPU / 16 GB, $0.60 an hour for 8 CPU / 32 GB and $1.20 an hour for 16 CPU / 64 GB. Asleep is free. Each plan sets how many computers it keeps, how much memory can be awake at once and the largest machine.

### Do I need Caveman Cloud to use Caveman Code?

No. One account holds both, and each has its own plan: a Caveman Code plan for the coding agent, and a Cloud plan for the agents your team builds. Neither needs the other, and Code credits and Cloud usage are billed separately.

### Is Caveman really free?

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.

### Do I need a card for Free?

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.

### How is Team priced?

$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.

### How does the Team trial work?

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.

### How do Caveman Code gifts work?

Paying for a Caveman Code plan earns you free months of Pro to give to friends: one with Pro or Pro $60, two with Max $100, three with Max $200. Each gift is one month of Pro, free. Your friend adds a card to start, and nothing is charged for the gift. We'll email them three days before it ends. If they'd like to keep going, Pro continues at $20 a month on that card; if not, they cancel before then and pay nothing. Gifts are for accounts that have never had a Caveman Code plan, and each card gets one free month.

### How is Enterprise priced?

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

### What happens at my billing limit?

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.

### What counts as an event?

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.

### What does Free share with Caveman?

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.

### Do you mark up model tokens?

No. On Cloud, 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. On Caveman Code plans, credits are spent at each model's list price.

### Will my bill drop by these percentages?

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.

### Does my code leave my machine?

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.

### What is the difference between the agent and the SDK?

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.

### How are the free tools licensed?

The skill, CLI, SDKs, middleware, extension and Browse are Apache-2.0. Releases before 3.0.0 keep the licence they shipped with. The standalone cavemem package is MIT. Caveman Code is proprietary, not open source. Check each repository for the full terms.

### Can Cloud run in our own cloud?

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.

[Explore every product](https://caveman.so/products)
