DeepSeek No Further

“A free and open internet is vital for the freedom of expression.” ― Oscar Auliq-Ice

May 09, 2026

A Whale of an Upgrade

We just pushed DeepSeek-V4-Pro, V4-Pro:web, V4-Flash, and V4-Flash:web live on app.mor.org + the inference marketplace.

Why this actually matters:

Frontier performance at open-source cost - - - These MoE beasts rival (or beat) top closed models on agentic coding, long-context reasoning (1M tokens), math/STEM, and knowledge - - - but you pay pennies on the dollar.
No more vendor lock-in or surprise bills - - - One-line OpenAI-compatible switch to decentralized routing. TEE-backed providers mean hardware-attested integrity. Your app runs on infrastructure nobody can shut down.
Instant access + propagation - - - Live now on app.mor.org, flowing to the full marketplace. Providers earn MOR for serving it; you get reliable, cheap, uncensored inference.

AI & DeAI Alpha

Killer robots? What about… Italian robots? VeniceAI is partnering with Strike Robot to train them in the decentralized ways of the Force.

Strike Robot@StrikeRobot_ai

IT'S OFFICIAL! We're announcing our partnership with @AskVenice — a privacy-first, uncensored AI inference platform on @base — as a primary inference API backend for StrikeRobot’s product offerings. The partnership involves Venice's credit sponsorship and co-development with

2:00 AM · May 7, 2026 · 32K Views

40 Replies · 32 Reposts · 191 Likes

Tokens are burning faster than 1600s London. Alex Whedon’s new LLM might be the solution. Warning: since it doesn’t analyze every word possibility, incoherency is going to be more damaging than ever.

Alexander Whedon@alex_whedon

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens -

2:00 PM · May 5, 2026 · 12M Views

1.44K Replies · 2.85K Reposts · 22.5K Likes

Meanwhile, Mistral AI is killing the SWE-Bench…

Further, arxiv discusses: Are your tools killing your session?

And on Reddit, we have the top GithHub searches for this month. (Great for anyone looking to optimize their GitHub visibility)…

Lastly, a new model release from the open-source community!

Zyphra@ZyphraAI

Today we're releasing ZAYA1-8B, a reasoning MoE trained on @AMD and optimized for intelligence density. With <1B active params, it outperforms open-weight models many times its size on math and reasoning, closing in on DeepSeek-V3.2 and GPT-5-High with test-time compute. 🧵

7:10 PM · May 6, 2026 · 845K Views

59 Replies · 1.27K Reposts · 1.68K Likes

Podcasts & Media

Morphean Bobski quietly launched the Own Your Mind podcast. He’s got videos on everything from DeAI’s importance to AI’s holding wallets.

Most of our contributors are at Consensus, but they tuned in for a short discussion (some of them on-site). You can still get your dose of DeAI news from our spaces.

Meta of the Week on X

Elon Musk is going to save us all from both planetary destruction and damaged rate limits. He’s rocketing the entire industry to success… pun intended.