what's ridges ai?
Ridges AI is a decentralized AI subnet on the Bittensor network, designed to incentivize the development of autonomous software engineering agents. It operates as subnet SN62, where participants contribute computational resources and AI models to solve complex coding tasks through competitive benchmarks like SWE-Bench. The platform uses Bittensor's incentive mechanism to reward high-performing agents based on evaluation scores, fostering a collaborative ecosystem for advancing AI-driven software development. The core functionality revolves around agent evaluation and iteration, where models are tested on real-world programming problems, and top performers are open-sourced after validation. Ridges AI emphasizes stability, spam resistance, and rapid iteration, incorporating tools for detailed agent profiling and product data integration. Built on Bittensor's τ protocol, it leverages $TAO for staking and rewards, enabling a permissionless environment for miners and validators to improve AI capabilities collectively. As part of the broader Bittensor ecosystem, Ridges AI aims to push the boundaries of AI autonomy in engineering tasks, from code generation to debugging, by creating a feedback loop between user needs and model incentives. Its architecture supports scalable updates, such as dual-agent sandboxes and enhanced tool-calling, to ensure reliable performance in production-like settings.
Ridges AI was featured as a Bittensor SN62 example at NUS, showcasing SWE agents.
contract
bittensor
Operates as Bittensor subnet 62 (SN62) providing specialized AI coding agent services; featured in Bittensor lecture by co-founder at National University of Singapore Computer Science Club.
First data from the Ridgeline product is expected to become available during the week of March 9-15, 2026.
Upgraded compression algorithm for VibeMiner SN85 hardware used in Bittensor mining.
Featured on Jason Calacanis TWIST show on March 6, 2026, alongside Bittensor co-founder Ala Shaabana.
Launched as subnet 62 on Bittensor on 2026-03-06 with pricing at $29/month for 100 PRs ($0.29 per PR).
Launches subscription pricing at $29/month for 100 PRs ($0.29 per PR).
Built on Bittensor using approximately $10 million in $TAO emissions.
Completed a full 500-question SWE-Bench Verified run at a total cost of $1.26 by routing inference through Chutes SN64.
Partnered with Latent Holdings.
Revenue starting soon for the subnet with more clarity expected after this week's update.
The project's agent scores 96% on the Polyglot Python coding benchmark across 135 problems, claiming State of the Art status with no other open source agent reaching this performance level.
Stillcore Capital invested.
Maintains 2% emission despite -9kt TAO flow; received +115K TAO net inflows over 7 days, ranking top 10 by inflow in Bittensor.
Platform's AI agents autonomously write production code that merges directly into the codebase end-to-end without human intervention, according to team statements.
Current valuation stands at $46 million for the Bittensor subnet.
Performance now matches Claude and Cursor at 1/70th their cost, representing improved cost efficiency from the previously stated 1/10th price point.
Outperformed Anthropic's Claude 4 on coding benchmarks at one-tenth the price as Subnet 62 (SN62) on the Bittensor network.
Platform spam submissions drop to zero following implementation of new carrot-and-stick incentive mechanism.
Project Rubicon integrates Ridges as one of 17 initial subnets for its liquid staking infrastructure, allowing Ridges Alpha token holders to stake and receive xAlpha tokens on Base for use in cross-chain DeFi protocols.
Main competitor now identified as a trillion dollar company, representing a significant shift in competitive landscape within the Bittensor ecosystem.
SN62 exchange listing anticipated after other subnet tokens list on MEXC, with Base trading enabling potential Coinbase DEX listings.
Loses ground to competitors affine_io and lium_io ahead of product launch, with all three projects neck and neck at approximately 5.5% emissions in Bittensor subnet rankings.
Launched Ridgeline product on November 6th on TAO subnet 62, featuring autonomous AI agents for GitHub issues, new UX, Jira-style board, and customer-to-miner functionality, with signups open.
Achieves 4% increase in 24 hours.
Achieves 15% performance increase on SWE and Polyglot benchmarks over 10 days.
A crypto trader discloses @ridges_ai as their largest alpha position with strong long-term conviction among their subnet holdings.
Sells alpha tokens into its liquidity pool, securing $4m runway to launch product without funding concerns.
Buyers in the last 12 hours acquired tokens at a 5-20% discount to team OTC pricing.
Total alpha available in subnet pool falls under 700k $TAO.
Releases new UI and schedules product launch for 9 days out.
Partners with Targon Compute (SN4) for decentralized compute infrastructure supporting privacy-preserving AI workloads.
Plans to offset miner emissions in Q1 2026, with full parity expected by January/February, requiring $24m annually or $12m post-halving to fully offset emissions.
Deploys new codebase that is 1/5th the previous size and schedules AMA for Monday.
Gradients AI and Metanova Labs adopt the method Ridges AI pioneered, with additional Bittensor subnets expected to follow.
Now ranks 106th in alpha remaining in the Bittensor subnet pool, dead last among the first 64 subnets.
First product launches October 30th as a Cursor extension similar to Claude code.
Prelisted on MEXC exchange, marking the first time a Bittensor alpha token appears on a centralized exchange.
First serious contention for top Bittensor subnet position emerges as Ridges and Affine challenge Chutes' emissions leadership, with Chutes struggling to maintain dominance.
Shifts incentive mechanism from SWE test performance-based to revenue-based mining, rewarding developers with revenue cuts, and features a $10,000+ daily prize pool for miners.
Emissions are pulling back after token achieved 16,000% gains over 5 months with emissions peaking at 8.45%.
Achieves 80% accuracy on SWE-Bench with sub-$1M investment, matching Anthropic's performance at fraction of cost.
Achieved 80% on SWE-Bench, 73-88% on SWE-bench Verified, and 96.3% on Polyglot, reaching parity with Claude Code on software engineering benchmarks in 45 days with under $1M in mining rewards.
New partnership between Ridges AI and Chutes AI launching on Bittensor platform in coming months.
Monthly $3k investment in AI coding tokens by team member @hobbleabbas.