.@near_ai’s confidential models are now available through @say_gm_ on Bittensor. GM is becoming a gateway for private AI across ecosystems.

The Hippius mobile app is live on Android. Your photos and files, backed up automatically and end-to-end encrypted on your phone before they ever upload. The same account and storage you already use on desktop and web. Your whole cloud, now in your pocket. iOS is coming next.

The most dangerous mind in history is the one aligned to a state.

Novelty Search :: Subnet 105 :: @b1m_ai Beam is building programmable bandwidth on Bittensor, moving data across clouds, storage, P2P networks and AI infrastructure. Six months after launching on mainnet, Beam is already processing ~1 TB every 30 minutes, with the next phase focused on real-world adoption and larger-scale workloads. Live on Thursday :: 9PM UTC :: via Bittensor Discord Hosted by @const_reborn

The future of cybersecurity will be dictated by a defender's ability to not only ingest unique attacks, but generate the strains that don't exist yet and ship the detection ahead of them. As the rest of the industry degrades, the largest financial institutions will consolidate on the single provider that can do this at scale. That provider needs a constant supply of unique attacks to work from, and a subnet is the only structure that pays a swarm of attackers to supply them. bittensor:native #SN61

Steadily building the most decentralized GPU cloud Lium now has capacity from 68 datacenters across 21 countries Have GPUs? Join now. Lium pays you even for idle minutes. Make your nodes rentable in 5 minutes -> docs.lium.io/providers/quick…

SWARM just landed its second commercial project. Mevi Dron picked us to build an avalanche-prevention drone, starting with Andorra's biggest ski resort. The mission: make controlled avalanche release safer and more efficient. The SWARM lab is at work. More below 👇

RedTeam is working toward integrations with three banking companies in the top 10% of the Fortune 500. On the latest Novelty Search, @_redteam_ explains how #SN61 miners strengthen Innerworks’ cybersecurity products. They also demo their Immune System, where AI agents created three new attacks from a miner’s submission and wrote the code to stop all three. Hosted by @const_reborn Full episode in the first comment

Dashboard is a quick work in progress, but for visibility into a run here ya go! parallax.chutes.ai/ This is around $11/b tokens, insane actually. MFU also insane. The whole thing, pretty legendary, and inference... Using a few nodes from @lium_io also!

New Hippius Drive plans are live. Small monthly price, serious storage. Free ($0): 10GB Starter ($4): 500GB Plus ($7): 2TB Max ($22): 10TB Scale ($50): 25TB End-to-end encrypted. Start free, upgrade whenever you need more space.

We're excited to announce that we're going to be joining Y Combinator in Fall 2026. The team is super pumped to be working with @golda and co to continue on our journey of creating the best in class open source models for agentic commerce.

We just ran Qwen3.6 35B at 14,499 tokens per second. on 1 lium GPU. 85% cheaper than Openrouter. how you can do it too ⬇️

Novelty Search :: Subnet 61 :: @_redteam_ RedTeam turns Bittensor miners into a 24/7 network of ethical hackers. We’ll discuss how SN61 uses that competition to stay ahead of evolving bot attacks, its five new enterprise pilots, and what comes next with RedTeam’s upcoming Immune System. Live on Thursday :: 9PM UTC :: on X + YouTube Hosted by @const_reborn

Can decentralized AI compete with Big Tech? @const_reborn, co-founder of @bittensor, joins @2084Richardson to discuss: - Bitcoin-inspired AI - TAO & subnets - Decentralized AI - Intelligence markets - Open AI Would you trust AI more if anyone could help build it? Dive deeper🔽

@NVIDIAAI and CrowdStrike ran a red agent against a blue agent inside an isolated replica of NVIDIA's infrastructure and tested the result on eight unseen attacks. The loop is the right idea, however, the real constraint here is the quantity and diversity of attack generation. One harness, or one internal team, searches with one set of tools and priors, so it finds the gaps it was built to look for and the loop stops when it runs out of paths. The attackers a platform actually faces are a spread of people with different tooling, different skill levels and different reasons to be there, and the gaps they find are the ones that actually get used. Building a decentralised community of attackers is the key solve. Our miners are paid to get past the detection stack with any method they choose, which right now means over 100 attacks a day, all being automatically reverse engineered, strengthening our core cybersecurity models continually. bittensor:native

GPT-6 Astra is live on Gm. OpenAI’s smartest model yet. Tell it what you need done. It opens the browser, writes the code, runs the terminal, ships the result. Routed through a verifiable TEE, at 8.3% discount on SayGm. saygm.com/models/gpt-6-astra

United against the divided. Intelligence knows neither borders nor color. Uphold the torch with us to bring light where it is needed most. Open reasoning for humanity. Join the thousand-year intelligence federation. affine.io

"Point your coding agent at this repo and describe a game. It generates its own 3D assets, verifies them, and loops with a harsh critic until the result holds up." github.com/404-repo/404-game…

NVIDIA's own researchers estimate that 40 to 70% of agent LLM calls could be offloaded to small language models. Most of an agent's loop is narrow, repeatable work. Routing a request, extracting a field, formatting an output, calling a tool. That is small-model work at small-model prices. Intelligence per dollar is the metric we build for.

Built this game with glm-5.3 on engy.ai in just a few rounds of prompts. It's impressive how playable and complete it is. Open weights are moving too fast.