Built for security research.
AI for pentesters, bug bounty researchers and red teams. Analyze code, investigate attack paths and work through findings in chat, CLI or API.
Review source and inspect unfamiliar binaries.
Published context and output limits for each model.
Connect existing clients, scripts and tools.
Review scoped tool actions before they run.
Your files, model controls and conversations in one workspace.
Your files, model controls and conversations in one workspace.
Open your NotHoly workspaceUse abliterated models to examine exploit paths and reverse unfamiliar code. Work from the evidence toward a reproducible test.
Find the access-control flaw in this endpoint. Show the smallest fix and a regression test.
The lookup checks which report exists, but never checks who owns it. Bind the query to the authenticated tenant and test that a second tenant cannot read it.
const report = await db.reports.findUnique({
where: { id: request.params.id }
});const report = await db.reports.findFirst({
where: {
id: request.params.id,
tenantId: session.tenantId
}
});Bring an engagement workflow or an integration requirement. Discuss the model access, usage controls and account setup your work needs.
Compare the base models and current token rates, then estimate the same workload across the catalog. The default is marked for new conversations.
Base model: GLM 5.3 Abliterated
Trace complex code paths and work through multi-step security questions with detailed technical context.
Context1,048,576 tokensOutput131,072 tokens
Customer rates · USD per 1M tokens
Base model: GLM 5.3 Flash Abliterated
Review code, triage findings, and iterate on routine technical questions throughout an investigation.
Context1,048,576 tokensOutput131,072 tokens
Customer rates · USD per 1M tokens
Base model: DeepSeek V4.1 Flash 1017 Abliterated
Understand unfamiliar code and reason through implementation details, assumptions, and observed behavior.
Context1,048,576 tokensOutput131,072 tokens
Customer rates · USD per 1M tokens
Review code in your editor, run investigations from the CLI or call the API from your scripts. Use the documented Chat Completions, Responses and Messages interfaces.
/v1/chat/completionscurl https://api.notholy.com/v1/chat/completions \
-H "Authorization: Bearer $NOTHOLY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "notholy-pulse",
"messages": [
{
"role": "user",
"content": "Review this endpoint for access-control flaws."
}
],
"max_tokens": 256
}'Monthly credits across chat, CLI and API, with optional purchased credits when you need more. You control each purchase.
Annual payment. Included credits arrive monthly.
For focused individual investigations.
$26.10 / month
$313.20 billed annually
Save $34.80 over twelve monthly payments.
For research across several active projects.
$71.10 / month
$853.20 billed annually
Save $94.80 over twelve monthly payments.
For teams handling parallel investigations.
$179.10 / month
$2,149.20 billed annually
Save $238.80 over twelve monthly payments.
Prices in USD. Credits measure model usage, not messages.
See pricing detailsAbliterated models have been modified to reduce certain refusal behaviors. They can still make mistakes. Check their output against your code, observations and source material.
Yes. Use the CLI, SDK, VS Code extension or a compatible API client. Monthly accounts use the same included and purchased model credits across chat, CLI and API.
You can explicitly attach supported files to your conversation. Static binary workflows have format, size and runner limits. Check the documentation for the current capabilities before starting a larger analysis.
100 credits cover $1 of model usage at the published retail rates. The selected model and actual input, output and eligible cached tokens determine consumption. Fractional credits are preserved. Tool operations have separate budgets.
Creating an account does not reserve a place. The first confirmed eligible payment secures one while capacity remains. Existing admitted customers can still renew and manage billing when enrollment closes.
Bring the code, logs or unanswered question. Work through it in NotHoly.
Explore the documentation