Ridges (SN62) Competition 22 (May 14-19) Wrapped
Ridges AI wrapped Competition 22 on May 19, 2026, with a result set that doubles down on the subnet’s selective approach to emissions. Out of 419 agents created between May 14 and May 19, only 9 earned approval for emissions. The cycle also introduced a hard cost ceiling, reshaping how miners compete on Ridges (SN62).
How agents moved through the evaluation funnel
The Ridges (SN62) Competition filters submissions through a multi-stage pipeline, and Competition 22 showed sharp drop-offs after the early checks. Of the 419 agents created during the window, 405 passed pre-screening, a clearance rate of 96.7%. From there, only 155 agents reached beyond Screener 1, narrowing the field to 37.0% of the pre-screened pool. The validator stage cut deeper still, with 112 agents completing it at a 26.7% pass rate. At the final gate, just 9 agents earned approval for emissions, representing 2.1% of completed validator submissions.
A total of 236 unique miners submitted agents during the cycle. The best score reached 76.67%, while the average final score landed at 57.56%. That spread underlines how wide the gap remains between top performers and the broader field of submissions.
The new $0.29 cost ceiling
The biggest structural shift this round was the introduction of a maximum cost of $0.29 per problem. Every competing agent now has to solve problems inside that ceiling, or it drops out of the pipeline entirely. Actual numbers came in well below the cap, with the average cost per problem solved sitting at $0.10. That leaves a margin of $0.19, or roughly 65%, between what miners spend to solve a problem and what the cap allows.
According to the Ridges team, that gap matters because it translates directly into miner profitability once agents start serving real clients. As they put it,
When we deploy miner agents to serve real clients, miners are guaranteed to turn a profit.
Standards over participation
The framing from Ridges AI stays consistent with the subnet’s broader philosophy. Rather than rewarding every submission that clears a minimum bar, emissions flow only to agents that meet a sharper standard of accuracy and cost efficiency. The team put it bluntly: “Aligning incentives properly matters more than inflating participation numbers.”
That choice carries trade-offs. Fewer miners earn emissions per cycle, yet those who do operate at a level designed to hold up under real-world deployment conditions. For builders working on Ridges (SN62), the message reads straightforward, with standards climbing and emissions earned rather than handed out.
What to watch next
Competition 22 sets a new benchmark for cost discipline alongside accuracy. Upcoming rounds will likely test whether agents can push the 76.67% best score even higher while staying inside (or under) the $0.29 ceiling. With 236 unique miners competing this cycle, the participation base remains healthy even as the approval rate tightens at the top of the funnel.
FAQ
What changed in Ridges (SN62) Competition 22 compared to previous rounds?
Competition 22 introduced a $0.29 maximum cost per problem. Agents that solve problems above that ceiling drop out of the pipeline, turning the Ridges (SN62) Competition into a combined accuracy and efficiency contest rather than a pure accuracy race.
The evaluation funnel filters hard at every stage. Pre-screening cleared 405 agents, Screener 1 cut the field to 155, the validator stage left 112, and only 9 (2.1%) earned final approval. The design reflects the team’s stated priority of rewarding quality over participation.
With the $0.29 cap and average actual spend at $0.10, miners operate inside a 65% margin. According to the Ridges team, that gap is what makes approved agents profitable once they start serving paying clients in production.
