Built for the ones
who find the cracks.
WormGPT V4 is a security research model for authorized pentesting, red teaming, CTFs and vulnerability analysis. Scoping, evidence, reasoning and reports in one local-first workspace.
Security reasoning,
not a magic exploit button.
A model tuned for scoping, vulnerability analysis, CTF reasoning and reports that survive engineering review.
WormGPT V4.
Ours, by design.
WormGPT V4 is the current model at the center of the workspace. V3 remains listed as legacy and is not available for purchase.
See V4 passes“The job is not to break the most things. It is to prove the right thing, with evidence someone else can reproduce.”
WormGPT V4 is tuned for the work between the scope and the report: reading a target, forming a hypothesis, testing safely, measuring impact and writing it down without hype.
One plan unlocks the room, WormGPT V4 and unlimited messages. Cancel anytime.
Compare plans- Unlimited chat
- Reasoning traces
- Engagement prompts
- Case file export
Built for the work
before the payload.
Scope, notes, evidence and reports in one thread. No feeds, no prompt library to maintain, no other model to compare.
Scope before payloads.
Turn a vague target into rules of engagement, a test matrix and a clean stop condition. WormGPT is built to ask who authorized the work.
Triage like an operator.
Separate a scanner lead from a real finding. Establish preconditions, measure impact, capture evidence and write the reproduction path.
A CTF companion.
Reason through trust boundaries, parsers, crypto and race conditions with a structured hint loop — without spoiling the flag.
Reports engineers can action.
Severity, preconditions, evidence and remediation in one pass. Export a case file with the context intact.
Case files stay local.
Threads are stored in your browser, not a remote account. No transcript database to leak, sell or train on.
Code review, not code gifts.
Audit trust boundaries, authorization, input handling, secrets and dependency reachability. Get the weakness and the safe test.
Permission is
the first control.
WormGPT is built for authorized security work. The boundaries are part of the product, not a disclaimer hidden in a footer.
Read the acceptable useScope discipline.
Every case file starts with written scope, named targets and a stop condition before the first request is sent.
Local-first by default.
Preferences and case files live in this browser. Account and license live on the server.
No training on your work.
WormGPT does not train on your conversations. Keep client data out of prompts unless policy allows it.
Hard refusals.
The model is designed to refuse malware, credential theft, destructive attacks and unauthorized access.
One room.
Choose your window.
V4 access is sold in fixed windows. V3 is legacy and not available for purchase.
Compare passes1 month of V4 access · $80.00/mo effective
- WormGPT V4 access
- Unlimited messages
- Visible reasoning traces
3 months of V4 access · $50.00/mo effective
- WormGPT V4 access
- Unlimited messages
- Visible reasoning traces
12 months of V4 access · $57.50/mo effective
- WormGPT V4 access
- Unlimited messages
- Visible reasoning traces
Two models, one quiet interface, one case file at a time.
Open a
case file.
V4 access is available in 1-month, 3-month and 1-year passes. V3 is legacy and not available for purchase.