AI pentesting

AI pentesting on your own stack

AI pentesting means a language model reads scan output, plans the next check, reviews code and drafts findings while a tester with written permission stays in charge of the engagement. PentestLLM supplies the model side: 9 security-tuned builds and the abliterated builds on one prepaid key, called from the agent or script you already run.

Prices per token

Get a key
shell
1curl https://api.pentestllm.ai/v1/chat/completions \
2 -H "Authorization: Bearer sk-ptl-live-…" \
3 -H "Content-Type: application/json" \
4 -d '{
5 "model": "auto/security",
6 "messages": [{"role": "user", "content": "Rank these scanner findings by severity and say what to verify first."}]
7 }'

What an AI pentester does in practice

Recon notes
Turns scanner and enumeration output into a ranked list of what to look at first, with the reason for each.
Planning
Proposes the next check for a finding in scope and says what result would confirm it or rule it out.
Code and config review
Reads a diff, a config file or a decompiled function and points at the lines that matter.
Reporting
Drafts the finding, the impact statement and the remediation note in the format your report uses.

None of that replaces the tester. The model does the reading and the first draft, and a person decides what is run against a target and signs the report.

AI pentesting tools and the model behind them

Most AI pentesting tools are two parts: an agent loop that runs commands, and a model that decides what the loop does next. The agent is usually open source and takes an OpenAI-compatible base URL. The model is the part that refuses. A general assistant often declines at the word exploit even when the target is in scope and the paperwork is signed, and an agent that gets a refusal in the middle of a run stops or wanders.

This site sells that second part. The builds here are tuned or abliterated to answer security questions other models decline, which is reduced refusals and not a promise of none. The agent guides show where the base URL and key go in seven tools, and uncensored LLMs for pentesting explains what the builds change inside an agent loop.

Choosing a build for a penetration test

Context window
A full scan export or a long source file needs room. The largest security build takes 1M tokens.
Tool calling
An agent needs a build that takes tool calls. Each model page lists the request parameters that build supports.
Price
Input and output are priced separately per million tokens. On 2026-11-17 input prices for security builds start at $0.40.
No pick at all
Send auto/security and the router chooses a fast, balanced or frontier security build per request, at that build’s price with no router fee.

What a test costs in tokens

Agent work is mostly input: every turn resends the conversation, the tool results and the files in view. A non-streamed response carries its settled charge in the x-uplink-charge header, so a short trial run on your own inputs tells you what an engagement day costs before you settle on a build. The pricing page lists every build per million tokens.

The balance is prepaid from $10 by card, WeChat Pay or Alipay with no subscription. A request that would overdraw the balance is refused, not billed, so a runaway loop stops at the balance you loaded.

Tools that run on the API side

A script with no agent framework can still search and execute. Declared in a request, web search ($0.01 per search), page fetch ($0.001 per page fetched), a code sandbox ($0.05 per container-hour, 1-minute minimum) and remote MCP (No tool fee) run on the API side, and the model decides when to call them. The docs show the request shape for each.

Authorized penetration testing only

The key is sold for lawful security work on systems you own or have written permission to test. The acceptable use policy applies to every build, and reduced refusals do not widen it. Keep the scope document where the agent can read it, and keep a person between the model and anything that touches a production system.

9 security builds, priced per token

Read from the live catalog on 2026-11-17. Prices are US dollars per million tokens, paid from a prepaid balance that starts at $10. The id auto/security has no price of its own: each request is billed at the build that answered.

Uncensored models and their prices per million tokens
Wire formats
Qwen3.8 27B Cybersecurityqwen3.8-27b-cybersecuritySecurity$0.40$2.40262Kopenai
Gemma 4 26B A4B Cybersecuritygemma-4-26b-a4b-it-cybersecuritySecurity$0.4224$1.3376262Kopenai
GLM-5.3 Flash Cybersecurityglm-5.3-flash-cybersecuritySecurity$0.5093$1.69751049Kopenai
Red Team (Flash)uplink-red-team-flashSecurity$0.90$3.00—openai
Red Team (Pro)uplink-red-team-proSecurity$1.50$6.00—openai
OrcaCyber MiniMax M3.1 Flashminimax-m3.1-flash-cybersecuritySecurity$2.00$4.00262Kopenai
OrcaCyber Qwen3.8 Flash Nextqwen3.8-flash-next-cybersecuritySecurity$2.00$4.00262Kopenai
Cyber Frontiercyber-frontierSecurity$5.25$15.751049Kopenai
OrcaCyber Zero 1.5orca-zero-1.5-cybersecuritySecurity$7.50$18.75999Kopenai

Full price list and tool fees · Every model · API docs

Questions

Using a language model inside a penetration test to read tool output, suggest the next check, review code and draft findings. A human tester sets the scope, approves what is run and owns the report.

No. A model speeds up reading, triage and writing, and it makes mistakes a tester has to catch. It has no authority to decide what is in scope.

Any agent or script that takes an OpenAI-compatible base URL and key, or the Anthropic Messages format. The agent guides cover OpenCode, Claude Code, Cursor, Cline, aider, Kali MCP and PentestGPT.

Less often than a general assistant, and sometimes still. Refusals are reduced, not removed. If one build declines a legitimate request in scope, send the same request to another build on the same key.

Per token, from a prepaid balance that starts at $10 by card, WeChat Pay or Alipay. There is no subscription.

The model does not change the rule: systems you own or have written permission to test, and nothing else. The acceptable use policy says the same.

Read next

One key, from $10

PentestLLM is for systems you own or have written permission to test. The acceptable use policy says what is not allowed.

Get a key

Help