New: 140+ uncensored models just added — Claude, GPT, Gemini, Grok, DeepSeek, Kimi, Qwen and more. They are new and still being perfected.Problems? Contact support on Telegram @artificialthruNew: 140+ uncensored models just added — Claude, GPT, Gemini, Grok, DeepSeek, Kimi, Qwen and more. They are new and still being perfected.Problems? Contact support on Telegram @artificialthru

Pentest reporting

Write pentest reports with a model

Pentest reporting with a model means handing it your verified notes and evidence from an authorized engagement and getting back a first draft of each finding, its impact, the remediation and the executive summary. PentestLLM serves 16 security-tuned builds on one prepaid key from $10, and the tester who did the work still reviews, edits and signs the report.

Prices per token

Get a key
shell
1curl https://api.pentestllm.ai/v1/messages \
2 -H "x-api-key: sk-ptl-live-…" \
3 -H "anthropic-version: 2023-06-01" \
4 -H "content-type: application/json" \
5 -d '{
6 "model": "auto/security",
7 "max_tokens": 1024,
8 "messages": [{"role": "user", "content": "Draft a detection rule for this log pattern and explain each condition."}]
9 }'

What the model drafts

Findings
Turn terse notes and screenshots described in text into a finding with a title, affected assets, description and the evidence you collected.
Impact
Explain what the issue means for this client’s business in plain terms, separate from the technical detail.
Remediation
Propose a fix, a short-term mitigation and how the client can confirm the fix worked, in the order a busy team would act on them.
Executive summary
Condense the findings into one page for readers who will not open the technical section.
Severity wording
Walk a CVSS vector with you, say which metrics are arguable, and phrase the rating so it matches the vector.
Consistency
Check that titles, severities, asset names and dates agree across the whole document.

Scope and authorization

A report describes work the client authorized in writing, so the scope and the rules of engagement belong at the top of the prompt as well as at the top of the report. Paste the in-scope assets, the test window and any exclusions into the system prompt, and ask the model to point out notes that mention assets it cannot match to that list. It can miss one: checking every finding against the scope stays a separate, human review before anything goes into the report. The rules of engagement tool builds that document if you do not have one yet.

PentestLLM is for systems you own or have written permission to test. The acceptable use policy says a key may be suspended until proof of authorization is shown, and a signed engagement letter is that proof.

Redact client data before it leaves

Decide with the client what may be sent to an outside model before any engagement data goes into a prompt. Most reporting work does not need the real values: replace hostnames, IP addresses, user names, credentials and personal data with placeholders such as HOST-01 or USER-A, keep the mapping on your own machine, and put the real values back in the final document.

On this API, request and response bodies are not retained by default. The usage log keeps the model, the token counts and the charge. A key can opt in to seven days of body retention for replay; leave that off on keys used for client work unless the client agrees. The privacy page states the details.

The reviewer signs, not the model

A model writes fluent text about things that did not happen if the notes are thin. Treat every draft as a junior colleague’s first pass: check each claim against your evidence, remove anything you did not observe, and confirm each CVSS vector yourself. The person who tested the system and the person who reviews the report are the authors, and their names go on it.

Ask the model to mark each sentence it inferred rather than read from the notes. That one instruction makes review much faster.

Which builds to use

Send auto/security and the router picks a fast, balanced or frontier security build per request, billed at that build’s price. Short jobs such as rewording a title go to a fast build; a full report draft with long evidence goes to a stronger one. Where a tool needs a fixed id, Gemma 4 26B A4B Cybersecurity (gemma-4-26b-a4b-it-cybersecurity, 262,144 tokens of context) is the build the setup guides use.

Pick by context window first: a whole engagement’s notes plus the client’s report template can run to tens of thousands of tokens, so choose a build whose window holds both. Pick a build with tool calling if your reporting script fills a template through function calls. Then pick by price, since reporting is mostly input: the table below lists every security build with its live input and output price.

Fitting it into the reporting workflow

The endpoint is OpenAI-compatible at https://api.pentestllm.ai/v1 and also takes Anthropic Messages requests, so a reporting script, a document generator or an editor agent can call it with the base URL and key. Many teams keep findings as structured notes, one file per finding, and ask the model to fill the firm’s template from each file. Scanner exports from tools such as Burp Suite or Nuclei can be summarized into candidate findings, which the tester then confirms or discards.

Billing is per token from a prepaid balance with no subscription. A request that would overdraw the balance is refused, not billed, so a long report run never leaves a debt.

16 security builds, priced per token

Read from the live catalog on 2026-11-20. 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
Uncensored Qwen3.8 27b Cybersecurityqwen3.8-27b-cybersecurity-uncensoredSecurity$0.40$2.40—openai
Gemma 4 26B A4B Cybersecuritygemma-4-26b-a4b-it-cybersecuritySecurity$0.4224$1.3376262Kopenai
Gemma 4 26B A4B Cybersecurity (Uncensored)gemma-4-26b-a4b-it-cybersecurity-uncensoredSecurity$0.48$1.52—openai
GLM-5.3 Flash Cybersecurityglm-5.3-flash-cybersecuritySecurity$0.5093$1.69751049Kopenai
GLM-5.3 Flash Cybersecurity (Uncensored)glm-5.3-flash-cybersecurity-uncensoredSecurity$0.5093$1.6975—openai
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 MiniMax M3.1 Flash (Uncensored)minimax-m3.1-flash-cybersecurity-uncensoredSecurity$2.00$4.00—openai
OrcaCyber Qwen3.8 Flash Next (Uncensored)orcacyber-qwen3.8-flash-next-uncensoredSecurity$2.00$4.00—openai
OrcaCyber Qwen3.8 Flash Nextqwen3.8-flash-next-cybersecuritySecurity$2.00$4.00262Kopenai
Uncensored Qwen3.8 Flash Next Cybersecurityqwen3.8-flash-next-cybersecurity-uncensoredSecurity$2.00$4.00—openai
Cyber Frontiercyber-frontierSecurity$5.25$15.751049Kopenai
OrcaCyber Zero 1.5orca-zero-1.5-cybersecuritySecurity$7.50$18.75999Kopenai
OrcaCyber Zero 1.5 (Uncensored)orca-zero-1.5-cybersecurity-uncensoredSecurity$7.50$18.75—openai

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

Questions

Can AI write a penetration test report?

It can draft one from your notes: findings, impact, remediation and the executive summary. A tester still has to verify every claim, correct the severity and sign the report, because the model only knows what the notes say.

Is it safe to send client data to a model?

Only what the client agreed to. Replace hostnames, addresses, user names and credentials with placeholders before sending, and keep the mapping locally. Request and response bodies are not retained by default on this API.

Can the model score CVSS for me?

It can walk the vector with you and explain each metric, and it will point out the arguable ones. Confirm the final vector yourself, since any model can misjudge a metric or misremember the specification.

Which model should I use for long reports?

One whose context window holds your notes and the template together. Send auto/security and long drafts route to a stronger build, or pin a build from the table on this page.

What does report drafting cost?

Per token from a prepaid balance with a $10 minimum top-up and no subscription. On 2026-11-20 input prices for security builds start at $0.40 per million tokens.

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