what's chutes ai?
Chutes AI is a decentralized, serverless AI compute platform built on Bittensor's subnet 64 (SN64). It enables users to deploy, scale, and run open-source large language models and AI tools in production, powering trillions of tokens processed monthly. The platform supports features like fiat and stablecoin payments, with all revenue auto-staked to its alpha token. It serves thousands of paying customers and millions of users, integrating models for chat, image generation, and video processing. Chutes focuses on decentralized GPU compute to meet growing AI inference demand.
Chutes AI released SN64 Alpha; speculation surrounds subnet's >72B model decentralized training.
Links
x.com/chutes_aiReleased SN64 Alpha sharing on 2026-04-13.
Begins training a trillion parameter MoE model with novel decentralized methodology on April 20, 2026.
Processes 12-16% of OpenRouter's daily token volume.
Received +340K TAO in 7-day net inflow, ranking as the top subnet by inflow in the Bittensor ecosystem.
Partners with GenLayer to integrate decentralized LLM inference from Bittensor, enabling scalable, cost-efficient AI for validators and Intelligent Contracts.
partnered with Harvard
Recorded a 40% price increase over 7 days in the subnet economy alongside other Bittensor subnets.
Released end-to-end encryption (E2EE) for AI inference on public LLM models on March 12, 2026.
Launched Model Routing feature, enabling API request routing across up to 20 models via a single alias, and offers private inference aggregation service for $18-20 per month.
Has a subsidy ratio of 76%
Generates $5.5M annualized revenue from decentralized inference services and Bittensor's Subnet 64 (SN64) as of March 2026.
Mentioned as a hosting provider option for autonomous Bittensor mining infrastructure, alongside Basilica, in a community member's experimental setup description.
Generates $22K daily revenue, up from $12K.
Generated $220K revenue in February 2026 from paid subscriptions; generates estimated $1.3M-$2.4M annually from AI inference services.
Powers Pax Historia AI alternate history sandbox game, providing decentralized compute infrastructure, reached 35,000 daily users and millions of rounds played.
Operates auto-staking buyback on SN64 using inference revenue generated by Rayon Labs.
Integrated ParaClaw plugin for cheaper inference, enabling access to 60+ models in under 60 seconds.
Featured in Bittensor lecture by co-founder at National University of Singapore Computer Science Club as an SN64 (Serverless + TEE Compute) example, enabling confidential inference across 15+ frontier AI models with hardware-enforced privacy.
Generated $5.7M in revenue during 2025.
Announces collaboration with Affine team to monetize models through their platform, with technical integration supporting compatibility through Affinetes standard and basilica deployment.
No longer holds #1 subnet position on Bittensor, overtaken by Affine subnet after maintaining top ranking since dtao launch.
Partners with zeussubnet to integrate weather forecasting models, starting with Microsoft Aurora.
Reached $94M market capitalization, largest among listed Bittensor subnets.
Processes over 25 trillion total tokens, up from 21 trillion tokens reported in October 2025.
Integrated as initial subnet with Project Rubicon liquid staking infrastructure, enabling cross-chain staking of Alpha tokens and liquidity on Base through xAlpha token issuance.
MEXC exchange listing announced, broadening distribution within the TAO ecosystem.
Listed on MEXC exchange, second centralized exchange listing for a Bittensor subnet project.
Processes 120B+ tokens daily as compute subnet SN64 on Bittensor, providing decentralized inference services for AI agents, reaching 21 trillion total tokens processed.
Integrated into OpenRouter's default routing, ranks #1 open source inference provider, processing 40 billion tokens daily for MiniMax M2.5 model inference, performing strongly alongside Venice over 30 days.
Staking yield increases to 70%, up from 68% previously reported on October 15, with SoS prices at 1.00.
Bitcast (Bittensor subnet 93) integrates chutes ai for all YouTube analysis operations, reporting the solution is significantly cheaper than their previous setup.
Chutes (SN64) offers 68% staking yield on $TAO, ranking mid-tier among major subnets with 1.33M $TAO total staked across all subnets.
Facing first serious competition for #1 subnet position from ridges_ai and affine_io, struggling to maintain emissions dominance.
Plans announced to back up all images and potentially HuggingFace models on Hippius subnet, representing a technical partnership within the Bittensor ecosystem.
CEO floats idea of alpha token exchange listings with confirmed interest from several centralized exchanges.
Achieves 90% cost reduction versus Google Cloud while processing trillion-token scale.
Daily token production on OpenRouter exceeds Elon's model by 200%, showing significant market leadership.
Reached $100M market cap as the first subnet, with $760M fully diluted valuation, and ranked #2 among Bittensor Subnets (SN64) by market cap on CoinGecko.
Operates an inference marketplace as a #1 Bittensor subnet, offering 67% discounts on AI models subsidized by $TAO emissions, with sustainability concerns noted.
Reached 500,000 users, processes over 5M daily inference requests, and totaled 9.1T tokens processed. Holds $100M market cap, is largest free model provider on OpenRouter, and included in Bittensor subnets watchlist as SN64.
Fiat payment integration launches with revenue potentially funding token buybacks.
Templar I (1.2B parameter LLM) launches as first permissionless AI model trained on network, enabling no-code agent deployment through squad.ai platform.
Receives 518 TAO daily (14.4% of total network emissions), valued at $142,000 daily or $52 million annualized, ranking among 4 Bittensor subnets per Khala Research.
Daily token processing volume reaches 90B tokens, marking 28.6% increase from previous high.
Token processing volume increases to 90B daily, up 28.5% from previous high of 70B.
Competitive pressure intensifies as TAOHash narrows performance difference to 3% gap.
Active trader re-enters position after previous successful exit, signaling potential market reversal.
Processing 70B tokens daily (up from 55B) with 190K users (up from 166K) and 2K daily signups; ranks #1 in open source AI model provision.
Serverless AI compute platform sees 166k user spike alongside elevated trading activity.
Token serving volume shows exponential growth from 5B to 55B tokens daily between March and May.
Infrastructure milestone: 1.2 trillion tokens processed with 46 billion daily peak achieved.
Token emission rate of 2 per second with 700k circulating supply creates gap between $70M market cap and $2B FDV.
Major ranking shifts occur in Bittensor ecosystem as PTN exits Top 3 and Templar enters Top 5.