What the heck is Gamma? It's the biggest change to Bittensor's token design since dTAO? @const_reborn and @KeithSingery broke it down on day 2 at @ExploitSummit, the day after Const announced it on stage day 1. The 36 minute fireside chat is required viewing material, but I'll give you the cliff notes. > Gamma is a usage credit you mint by burning subnet Alpha, priced against a stable oracle basket, so $5 of Alpha becomes 5 Gamma that never changes in quantity once minted. > It is a one-way trip, since Gamma can't be converted back to Alpha and only the subnet owner key can accept it, which means 100% of the value capture flows through Alpha. > It puts the buyback-and-burn credit model that @lium_io, @chutes_ai and @TargonCompute already run on private databases directly on chain, with no email, no API key and one atomic transaction from any chain. > Subnet owners can redirect part of miner emissions to automatically buy other subnets' Gamma, so a product team can pull in inference, compute and storage from across Bittensor without raising outside capital.
Bittensor is now accessible from Ethereum. @taodotcom just provided access too the $TAO and subnet tokens through their new feature, crosschain USDC swaps from Ethereum Mainnet to Bittensor. You can swap $USDC or $ETH on Ethereum for native $TAO or subnet alpha tokens without moving funds to a centralized exchange as an intermediary bridge. Getting into Bittensor has always been the hard part, but thanks to @taodotcom and @LayerZero_Core Bittensor is now more accessible than ever.
Drug discovery is a search problem. There are more possible molecules than anyone can test, so pharma picks a target, searches a narrow slice of chemical space with one team and one method, and hopes. @metanova_labs built Bittensor Subnet 68 to search it a different way. Anyone can submit candidates against a specified target, submissions get scored and ranked, the best ones win, and NOVA pulls the strongest into its own pipeline. Small molecules, nanobodies and algorithms are live now. No lab and no pharma job required to enter. What it's searching for is Agency Therapeutics, a proposed framework for drugs that improve the biological capacity to regulate yourself. Like how GLP-1s reduce what you have to resist, ADHD meds increase the capacity to act, and psychedelics may help better patterns stick. @EliLillyandCo booked $36.5 billion on Mounjaro and Zepbound last year selling drugs that make hunger easier to resist. They were built for metabolism, and that effect was a byproduct. NOVA is designed to find Agency Therapeutics on purpose, and the first hits from that search are already at a lab being tested. Imagine what it means when NOVA productizes its first drug out of an open competition.
Add Erdős Problem 14 a 34 year old mathematics problem to the list of problems Bittensor SN66 @conjectures_io has solved by incentivizing miners. That is now 4 of the most difficult math problems in the world solved over the past week. You know, just a result that proves a square-root lower bound on exceptions to unique representation as a sum of two elements of any set of natural numbers. You know, just amassing a cracked mining pool to with a resume that looks like this: SN66 Miners, Mathematics Professionals > Solved Erdős Problem 14 > Solved Erdős Problem 196 > Solved Erdős Problem 96 > Solved Green's Problem 51 Probably nothing.
The market is starting to wake up to the fact that Bittensor SN114 @SomaSubnet can save you money on your inference. SOMA's context compression launched around 10% estimated token savings when using GitHub Copilot with DeepSeek V4 Pro. Now its estimated to be ~15% Miners have already improved SOMA Compressor by 50% since launch. SOMA's SN114 α token is up over 20% vs $TAO since 9/11
Context compression just landed in GitHub Copilot. SOMA (Bittensor SN114, @SomaSubnet) sits between your agent and DeepSeek V4 Pro and strips the context you're paying to re-send. Their product, Compressor for Copilot, shipped at ~10% savings and SOMA miners have already improved the compression model to achieve 15% savings. I hold SN114 alpha. Good product, and every team running agents is hunting for ways to cut token spend. Soma's compression model seems like the preferred method over having your agent talk like a caveman. @Gartner_inc forecasts AI model spend at $32.6b for 2026, up 110% year over year. Compression is easy to implement, and SOMA's architecture on Bittensor means miners continuously compete to improve the compressor. Compress. Pay less. @SomaSubnet.
Captured a good chunk of the morning down move on $SPCX Watching to see if ~$146 breaks. If it does looking for $141. https://t.co/ZgpGCyjgO7
Xavier spent his quant days hunting for optimal trading strategies. Now he's hunting for the optimal way to run a model. @xavi3rlu is speaking at @ExploitSummit on @pareton_ai, an open competition to make inference faster on the models we already have. That's a big opportunity. Efficiency gains hit every company running inference today. Looking forward to tuning in @ExploitSummit in from Montreal Sept 28-29
Kanye West and @green_compute_ (Bittensor SN110) actually share the same business model. > Waste stream > Captured > Delivered to an audience Instead of "stranded gas" @green_compute_ harnesses biogas from farm waste powering AI compute. It is a pretty straight forward business model, but they have teased an announcement that they will be making at the end of the month on stage at @ExploitSummit in Montreal. My guess as to what will be announced: > Expansion beyond biogas into other renewables Looking forward to tuning in to find out.