<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>SolanaLM Blog</title><description>SolanaLM is a decentralized, OpenAI-compatible AI inference network for the agent economy — pay per request in SOL with sub-second settlement, run nodes on cheap DePIN GPUs, with an on-chain audit trail heading toward verifiable inference. Self-hostable, MIT-licensed.</description><link>https://solanalm.cryptuon.com/</link><language>en-us</language><item><title>Running a SolanaLM node: real numbers from week one</title><link>https://solanalm.cryptuon.com/blog/running-a-node-economics-real-numbers/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/running-a-node-economics-real-numbers/</guid><description>Field notes from running an inference node for a week. Hardware, hours, payouts, and the operational reality that nobody puts in marketing copy.</description><pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate></item><item><title>The threat model for onion-routed LLM inference</title><link>https://solanalm.cryptuon.com/blog/onion-routed-inference-threat-model/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/onion-routed-inference-threat-model/</guid><description>Privacy-preserving inference sounds great in marketing copy. The honest engineering question is: what attacks does the 3-hop onion circuit actually stop, and what does it not?</description><pubDate>Tue, 12 May 2026 00:00:00 GMT</pubDate></item><item><title>OpenAI API compatibility is a distribution strategy, not an integration</title><link>https://solanalm.cryptuon.com/blog/openai-compatibility-as-distribution-strategy/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/openai-compatibility-as-distribution-strategy/</guid><description>Every decentralized inference network has to answer: how does an existing application start using us? Speaking OpenAI&apos;s schema is the cheapest, most honest answer.</description><pubDate>Tue, 28 Apr 2026 00:00:00 GMT</pubDate></item><item><title>FedAvg, FedProx, FedAdam, SCAFFOLD: what actually breaks in production federated learning</title><link>https://solanalm.cryptuon.com/blog/federated-learning-algorithms-in-practice/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/federated-learning-algorithms-in-practice/</guid><description>Four algorithms ship in the SolanaLM training runtime. They are not interchangeable. Here is what each one is for, what breaks when you pick wrong, and how to choose.</description><pubDate>Wed, 15 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Why we picked Solana for per-request AI inference payments</title><link>https://solanalm.cryptuon.com/blog/why-solana-for-ai-inference-payments/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/why-solana-for-ai-inference-payments/</guid><description>A 400ms finality target and ~$0.00025 per-transaction fee are not nice-to-haves for a decentralized inference market. They are load-bearing. Here is the engineering case.</description><pubDate>Thu, 02 Apr 2026 00:00:00 GMT</pubDate></item><item><title>SolanaLM vs Bittensor: a deep dive on settlement, schemas, and subnet doctrine</title><link>https://solanalm.cryptuon.com/blog/solanalm-vs-bittensor-deep-dive/</link><guid isPermaLink="true">https://solanalm.cryptuon.com/blog/solanalm-vs-bittensor-deep-dive/</guid><description>Bittensor pioneered decentralized AI markets. SolanaLM bets on a different chain, a different schema, and a different definition of done. Here is the honest engineering comparison.</description><pubDate>Sun, 22 Mar 2026 00:00:00 GMT</pubDate></item></channel></rss>