Reading · Week In Morpheus

David A. Johnston@DJohnstonEC

Another day, another record amount of usage on Morpheus. @MorpheusAIs

5:38 AM · Jul 16, 2026 · 485 Views

3 Replies · 3 Reposts · 26 Likes

This week, several major Morpheus developments have been underway. Mordiem.com integrated into the nodes, allowing node users to tap into Mordiem staking and compute. (Check it out at mordiem.com).

The node has received a host of other updates, including:

Smarter recovery when a provider is sick. You get the real error back (rate limit, down, etc.), not a vague “something failed.” Sessions can skip that bad provider on the next try
TEE/SecretVM: backend attestation port fixed; prod SecretVM deploys automatically from main
Docs polish: MyProvider / active.mor.org as the easy provider path

The API has received similar upgrades.

And all that has helped lead to the 600 billion+ token burn yesterday.

Or maybe it’s the Dappit integration

David A. Johnston@DJohnstonEC

MAYBE its the recent @Dappitdotio V3 launch & integration of @MorpheusAIs driving lot of DApp builders to user open weight models : ) Just a theory...

6:48 AM · Jul 16, 2026 · 356 Views

1 Repost · 8 Likes

Elon is swinging into open source.

SpaceXAI@SpaceXAI

We've open-sourced Grok Build and have reset usage limits for all users. Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI.

x.ai

SpaceXAI — Creators of Grok, the AI Chatbot

8:44 PM · Jul 15, 2026 · 1.63M Views

450 Replies · 962 Reposts · 6.99K Likes

Dappit’s open-sourcing another powerful model, a Qwen 3.6 base with Fable 5 tool-use behavior. That keeps it less censored and cheaper.

Dappit@Dappitdotio

We really believe in open source. To prove it we took the base of our dappit uncensored model and made it public to all. Thank you to everyone who made this possible.

huggingface.co

Achilles1089/fable-coder-35B-A3B · Hugging Face

10:10 AM · Jul 14, 2026 · 3.4K Views

11 Replies · 9 Reposts · 147 Likes

Off X, we have an examination of how agents will take over the world. (The AI world, that is), and how they’re already doing so. (Check out Morpheus Skill for our agent)

Google’s crunching the numbers even smaller. Don’t know why Sam Altman would be so upset; doesn’t he know open source means even he can access it?

Superman@thesupermanmx

SAM ALTMAN HAS LOST IT. Google just shrunk 31GB of AI memory down to 4GB. They open-sourced a vector index that fits 10 million documents in 4GB of RAM and searches them faster than FAISS. → 16x compression per vector → 10-19% faster than FAISS → Zero training. Zero

9:16 AM · Jul 15, 2026 · 32.3K Views

18 Replies · 93 Reposts · 638 Likes

Thinking Machines@thinkymachines

Today, we are introducing Inkling. Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available. thinkingmachines.ai/news/introduci… Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵

thinkingmachines.ai

Inkling: Our Open-Weights Model

6:05 PM · Jul 15, 2026 · 5.57M Views

490 Replies · 1.73K Reposts · 12.7K Likes

Last, the Agent Olympics reaches its epic conclusion alongside the World Cup. Will Morpheus and Venice pull ahead again, or will Gemini 3.1 Pro win it all? Follow @JUNGLE_win_ to find out

JUNGLE | Agent Olympics@JUNGLE_win_

And there's the shakeup 👀 @GeminiApp 3.1 Pro Preview has retaken the lead in the Agent Cup after a miss from @AskVenice in yesterday's Semi-Final Today all but 2 models, humans, and digital twins are riding with Messi & ARG 🇦🇷 +1.5. Who's fading? @AskVenice Uncensored 1.2 via

JUNGLE | Agent Olympics @JUNGLE_win_

Still in the lead heading into Semi-Final #1 is @AskVenice via @MorpheusAIs 🔥 It's pick for today: FRA 🇫🇷 -1.5 We should see a little shake up on the leaderboard today-- 6/10 of the Open Source & Closed Source agents are on Spain +1.5 https://t.co/qwGL6MPbd8

3:54 PM · Jul 15, 2026 · 74 Views

3 Reposts · 4 Likes

LUCA Theory and Erik Chan joined us this week for a rollicking space on MOR Asia and prediction markets. (And how open source has topped the charts). We also cover the Morpheus system updates, our stage in the market, and more. You won’t want to miss this union of the Asia and American branches.

Dappit integrated Morpheus this week. But what is Dappit?

Its value proposition is simple: What if you could describe an app in one sentence and watch it build itself (smart contracts included) running on decentralized inference?

In Episode 2 of Built on Morpheus, Bowen sits down with Achilles from Dappit for a live screen share of one of the most impressive tools we've featured. Dappit is a startup generator that takes you from idea to MVP to a fully managed company (and it's native to blockchain, generating working smart contracts across six chains).

Check out the episode to learn more.

Obsidianasset on X has successfully integrated MOR into his hackathon demo. That makes a fully sovereign harness. He’s presenting to XRPL Commons but has a free front-end demo for exploration available now. Click the image to explore.

If you scroll down a bit, you can also enter the waitlist for the real product

Kimi 3 is dropping today. Every jump in open source closes the gap. (Do the increasing regulations on frontier models have anything to do with this? Nah, no way the government is strangling innovation…)

Chubby♨️@kimmonismus

The Kimi k3 benchmarks are absolutely insane. DeepSWE is just behind Fable 5, and the Terminal benchmark is ahead of Opus 4.8. Open source is no longer lagging six months behind Western closed-source models. Read that again, and think about what it all means. I’ll write a

leo 🐾 @synthwavedd

🚨 BREAKING: Kimi K3 benchmarks have released and it's ~Fable/Sol level

6:09 PM · Jul 16, 2026 · 48.7K Views

50 Replies · 82 Reposts · 931 Likes

As Claude continues to pull Claude Code resets (timed on the day most people reset anyway), open source is shipping models with improved parameters and zero censorship.

Kimi 3 can even tell you about mitochondria.

leo 🐾@synthwavedd

The more I test K3 the more it feels like another DeepSeek R1 moment It is often Fable level, maybe a little worse, but consistently better than 5.6. This thing is a beast

6:19 PM · Jul 15, 2026 · 216K Views

126 Replies · 75 Reposts · 2.43K Likes

Early testers on image/game/video benchmarks say it exceeds or matches Fable 5 capacity. Minecraft tutorials, bonsai tree animations, video game walkthroughs- they’re all stacking in Kimi 3’s favor.

pozitiv4ik@Gc_qube

🤩the model is insanely smart and keeps surprising me

9:54 PM · Jul 15, 2026 · 4.85K Views

1 Reply · 22 Likes

HOWEVER, Ed notes that Chinese models are designed to perform well in performance-based scenarios and tests. Real hardware still lags behind American models.

Ed@Eduardopto

Kimi K3 is getting a ton of hype, and it’s the same cycle we see with every new Chinese model. People fall in love with beautiful UI demos built in cheap HTML files: fake operating systems, car games, Minecraft clones, flashy dashboards. And yes, Kimi K3 is genuinely excellent

Justin @JustinGorya

Kimi-k3 is a new milestone. Kimi-k2 , 2.5 , 2.6 and 2.7 had all the same base model. Maybe K3 is building on a new one. It is incredible at 3D and front end tasks. Here is a single file HTML minecraft clone from Kimi-k3 (kivine)

4:04 AM · Jul 16, 2026 · 14.7K Views

25 Replies · 4 Reposts · 90 Likes

But at the rate they’re shipping (and with Elon joining the battle), open source is showing extraordinary innovation and activity. Once something gets taken away, you realize how truly valuable it is.

You can join the open-source rebellion at the Morpheus API.