Messari.io
-
12:00 Apr 25, 2025
Rayon Labs operates a synergistic trio of Bittensor subnets—Chutes for serverless model hosting and inference, Gradients for decentralized AutoML fine-tuning, and Nineteen for ultra-fast, high-throughput inference—that together command roughly 29 % of daily TAO emissions. Chutes has exploded in popularity by offering headline open-source models like DeepSeek V3 for free via OpenRouter, amassing over 8 000 GPUs and delivering industry-leading latency, while Gradients’ competitive miner marketplace consistently outperforms major clouds on test-loss, making one-click training accessible to non-experts. Nineteen narrows its focus to a handful of models and rewards miners for raw throughput, resulting in some of the fastest free inference on the market. Rayon is building a freemium ecosystem—paid tiers on Chutes, enterprise-grade integrations, and an agent platform called Squad—to convert today’s subsidized usage into durable revenue and alpha-token buybacks. The strategy’s success, however, depends on weathering ...