Honest AI · cited · scored
An AI that cites its sources and scores its own confidence.
Straight Edge GPT verifies before it answers. Live search on anything checkable, primary sources ranked above summaries, a 0–100 confidence score on every reply, and a plain "I don't know" when the evidence isn't there.
Citations on every factual claim
Each claim links to the page it came from, and the page is actually opened and read before the claim is made — not inferred from a search snippet.
A confidence score you can act on
Every answer ends with 0–100 and a one-line basis. Above 80 means verified against sources it opened. Below 50 means it is reasoning, and it says so first.
Honest about what it doesn't know
"No evidence found" and "not enough data" are complete answers here. Low-confidence replies literally open with "I don't know the answer to that, but here's what I think…"
Live verification, shown in full
Anything checkable is checked against the live web at answer time. Every query it ran and every page it opened is visible, and the whole trace exports as JSON.
How an answer is built
- 01
Decide whether the question is checkable at all.
- 02
Search live, open the actual pages, rank them by source tier.
- 03
Cross-check claims across sources and log contradictions.
- 04
Score confidence, state the basis, and say what would change the answer.
Common questions
What makes an AI answer honest?
Three things: the claim is traceable to a source you can open, the model's certainty is stated numerically rather than implied by tone, and gaps are declared instead of filled with plausible text.
Can I use it without an account?
Yes. Three questions per network are free with no card and no signup. After that it's $19.99/month, or $9.99/month billed yearly at $119.88.
Does it work offline?
There's an offline knowledge mode that freezes it to training data with no network calls, for reasoning tasks where retrieval noise hurts more than it helps. Answers in that mode are labelled as unverified.
Three answers. No card.
Ask it something it could get wrong and watch what it does with the gap.
Start the free trial