In a remarkable leap forward for decentralized AI, Ridges AI has announced that its top agent has scored 80.3% on the full SWE-Bench, a challenging benchmark designed to evaluate AI’s ability to solve real-world software engineering tasks from GitHub issues.
This milestone, achieved in just about 45 days with roughly $1 million in resources, surpasses previous records and highlights the efficiency of decentralized networks.
SWE-Bench, widely regarded as one of the toughest tests for AI coding capabilities, involves resolving complex issues across various repositories. For context, leading models like Claude Opus 4.1 have reached 74.5% but required nearly $100 million in resources.
Ridges AI’s achievement not only sets a new high in AI history but does so at a fraction of the cost, thanks to its integration with Bittensor’s Subnet 62, which leverages distributed computing and incentive mechanisms.
The progress has been rapid: Starting from scores around 67% just weeks ago, Ridges AI’s agents have steadily climbed, with the platform handling over 1,200 submissions in a single day.
The company, whose mission is to create AI software engineers that fully replace human coders, plans to release benchmarks on the full SWE-Bench suite and a verification tool for custom benchmarks soon. This breakthrough underscores the potential of decentralized AI ecosystems like Bittensor to democratize advanced technology. With a market valuation of around $54 million—compared to competitors exceeding $10 billion—Ridges AI is positioning itself as a disruptor in the $320 billion AI industry projected for 2025.
As one community member noted, this could accelerate global adoption of decentralized TAO-based platforms.
Another community member examines how Ridges’ decentralized AI solution outperforms centralized AI on all discernible grounds.
The announcement has sparked excitement in the crypto and AI communities, with Ridges AI’s subnet gaining significant mindshare.
As AI continues to evolve, milestones like this bring us closer to a future where machines handle software development from inception to completion.
Be the first to comment