Independent audit · 6 services · Updated Jun 2026

Read the fine print.We already did.

Hidden behaviors, real limits, and policy changes the pricing page won't tell you — documented, sourced, and verified.

Hidden Behaviors
0
documented across all services
Services Audited
6
AI tools · design · coding
Last Verified
Jun 2026
audits updated within 48h of changes
Transparency Reports
6 audited · sorted by transparency
OFFICIAL company docs
OBSERVED community / benchmark
ESTIMATED not verified
Canva
Canva · Design · $15/mo
71% TRANSPARENT
3
Hidden Behaviors
AI credits & storage on separate reset cycles
No shrinkflation detected · most stable View report →
Claude
Anthropic · AI Chat · $20/mo
52% TRANSPARENT
5
Hidden Behaviors
Dual cap system — 5h AND weekly, either can block you
50% limit boost expires Jul 13, 2026 View report →
Perplexity
Perplexity AI · Research · $20/mo
44% TRANSPARENT
4
Hidden Behaviors
Deep Research cut ~99% — 500+/day → 20/month
Severe shrinkflation · price unchanged View report →
Cursor
Anysphere · Coding · $20/mo
41% TRANSPARENT
4
Hidden Behaviors
Overage billing ON by default — card charged silently
~40–60% request reduction since Jun 2025 View report →
ChatGPT
OpenAI · AI Chat · $20/mo
31% TRANSPARENT
5
Hidden Behaviors
No usage meter — limit discovered only on hit
Silent model downgrade on cap hit · $20 since 2023 View report →
Gemini
Google · AI Chat · $19.99/mo
28% TRANSPARENT
4
Hidden Behaviors
Fixed credits eliminated — replaced with opaque compute quota
No usage meter post May 2026 · lowest score View report →
What This Database Tracks
🔍
Hidden Behaviors
Policies and behaviors confirmed through docs, benchmarks, and community reports — that don't appear on the pricing page.
📉
Shrinkflation
When the price stays the same but what it buys shrinks — quietly. Every confirmed event is dated and sourced.
📊
Transparency Score
How much of each service's real behavior is disclosed upfront — weighted across limits, privacy, speed, and reliability.
Get alerted when policies change
We track silent limit cuts before most users notice them.
How we audit — full methodology, source classification, and formula →
↑