Kled V3 Launches on iOS, Rebuilding Its Solana-Paid AI Data App Around Lab

Kled AI released Kled V3 on 30 September 2026, a rebuilt version of its app that pays people to capture photos, video, audio and text for AI training datasets. The update, announced by Kled at 14:59 UTC and now live as version 3.0.0 on the iOS App Store, replaces the app's upload-first home screen with a ranked feed of paid tasks that AI labs specify. Contributors can still cash out to a Solana wallet, and the official Solana account described the app as "Powered by payments on Solana."

Kled pitches the release in the largest terms available. The company's announcement says it has "solved data collection for artificial general intelligence" and that "any consumer dataset that can exist" can now be collected "in under 72 hours." Co-founder and CEO Avi Patel called V3 "the data collection layer for agi." Those are Kled's claims, and the launch thread does not name a lab that has run a dataset through the new system yet.

Kled V3 replaces the camera-roll upload button with a ranked task feed

Kled V3's main change is the home screen. According to Kled's launch thread, the previous app was built around a single button for uploading photos already in a contributor's camera roll, while structured tasks sat in a separate tab most contributors rarely opened. That meant Kled "could only collect the data people already had, not the data labs actually needed," the company wrote.

V3 replaces that screen with a task discovery feed. Every live task is ranked per contributor, lab-priority tasks appear first, and the pay for each task is shown upfront. A recommendation system learns where each contributor's submissions get accepted, by modality, task type and region, and contributors can adjust the feed themselves. Kled's example of what this allows: if a lab needs 40,000 minutes of first-person video from Southeast Asia in a week, Kled can route contributors to that task directly. The feed works like a ride-hailing dispatcher, sending available supply to whatever job is posted.

108 templates across camera, microphone and screen recording

Kled V3 supports task types across image, video, audio, text and annotation, according to the thread. Kled's examples include video of someone reorganising a room, one object photographed from every angle, conversational speech in rare languages, flagging AI-generated images, and predicting the next button a person would press in an app. Each task is built from one of 108 templates spanning camera, microphone, screen recording, touch and keyboard input, and each template carries instructions plus examples of accepted and rejected submissions.

Kled says this lets it build any dataset a lab can specify into the app and deploy it to more than 500,000 contributors within 72 hours. The company also says it expects to let labs design and deploy tasks themselves from the same templates, which would make ordering a dataset self-serve. No date was given for that.

The rebuild took eight months, Kled said, after work with several frontier labs showed its existing infrastructure could not handle the tasks and data types they needed. Kled puts its network at more than 500,000 opt-in contributors across the US, Southeast Asia and Europe, collecting over 8 million datapoints a day across 38,000 structured datasets.

FD 1.0 fraud detection model screens every upload

Paying strangers for uploads invites duplicates, stolen images and AI-generated fakes. Kled V3 ships with FD 1.0, a fraud detection model the company says now runs on every upload. According to Kled, the model screens for duplicates, AI-generated images, content taken from the internet, edited files and behavioural fraud, and scores each submission for task relevance and quality before accepting it. By the company's own figures, FD 1.0 removes 98.4% of bad actors within their first 10 uploads. The thread also claims the model outperforms the best commercial systems on its internal evaluations; it has not published those evaluations.

How Kled contributors get paid on Solana

The V3 launch did not announce changes to payouts. The App Store listing says contributors can cash out "through PayPal, Venmo, or a Solana wallet" once they reach a $25 balance, typically within three to five business days, and that earnings depend on a submission's uniqueness, diversity, quality and current buyer demand. Kled's own token, KLED, also trades on Solana SOL$118.34+0.2%.

That payout rail is the Solana side of the story. Kled's contributors are spread across the US, Southeast Asia and Europe and earn small, variable amounts per task, and the Solana wallet option lets Kled send those payouts on-chain to any contributor who holds a wallet. The Solana account's comment on the launch points at the same layer.

On-chain, the KLED holder base had been shrinking before the release. Solana Compass token data shows wallets holding KLED fell from 14,220 on 3 July to 13,797 on 25 September, then rose to 14,208 by the start of 30 September, the day V3 shipped. Holder wallets are not the same as contributors, since many contributors cash out through PayPal or Venmo and never hold the token.

From a $12M data deal to Kled V3

Kled V3 follows the $12 million non-exclusive data licensing deal that we covered in June, when Kled said the agreement made its marketplace profitable. Kled's homepage lists $10 million in total funding and still describes its Android app as open for pre-registration; the V3 thread names only the iOS App Store and points AI labs to kled.ai/datasets to request data.

Kled competes for lab data budgets with established labeling and data companies such as Scale AI and Appen. Kled's pitch is that a phone app with half a million contributors, a template library and a Solana payout option can fill a custom order faster than they can. The first lab to publish results from a 72-hour order will show whether it does.