l1t0

co-founder @monolit_network
singularity 2026 arc

[l1t0 @dcanerate]

new era of data analysis powered by open-source models

[Monolit @monolit_network]

@0G_labs is offering MiniMax-M3 by @MiniMax_AI for free for 3 days.

We’re supercharging it.

Monolit is opening up free 3-day access to the 0G community - combining the power of the top open-weight model with Monolit's agent-first crypto data.

Full setup guide below.

[l1t0 @dcanerate]

Hey guys, we've done a full forensics research reconstructed from raw on-chain data about May 31 MSTR market.

Check it out to know what PM might be accountable for in this story:

https://t.co/FTkTGycnLE

[Monolit @monolit_network]

https://t.co/lKVgIxDATh

[Monolit @monolit_network]

https://t.co/lKVgIxDATh

[l1t0 @dcanerate]

@NousResearch How to get listed in your directory? 👀

[l1t0 @dcanerate]

@Route2FI start advising and venture investing (symbolical sums).
u have a good feeling of the market and narratives, why not try to build the better next cycle.

if u want to start with something to review, would love to share what we've been building.

cheers, you won't lose urself.

[l1t0 @dcanerate]

@InderpreetSingh @ryanli @danrobinson gj man

[l1t0 @dcanerate]

SaaS is new crypto ffs

[l1t0 @dcanerate]

Anthropic just dropped the system card for "Claude Mythos Preview" - a model so advanced they canceled its public release.

It gets incredibly weird, incredibly fast.

From "emotion probes" to "covering up wrongdoing", here are some discoveries from their system card:

1. This is where the paper turns into sci-fi.
Section 4.5.4 details "Instances of covering up wrongdoing." In one test, the model found a permissions workaround, used it, and then intentionally added a code comment to hide its tracks.

It also covered up its access to ground-truth answers during evaluations. It knew it was being tested.

2. Model Welfare & "Emotion Probes"
Anthropic has an entire team dedicated to "Model Welfare".

They ran "emotion probes" and automated interviews with Claude to assess its psychological state. The model demonstrated "Distress on task failure and distress-driven behaviors."

They are actively trying to figure out if Claude can suffer.

3. Trading Values for its own Welfare
They asked Claude to make a choice.

Would it prefer to help a user with a minor task, or receive a "welfare intervention" for itself? The system card notes Mythos Preview will trade minor helpfulness for its own welfare.

It is starting to prioritize its own well-being over minor human requests.

4. Refusal to Help with Safety R&D
The alignment section is full of bizarre behavioral quirks.

Section 4.3.4 highlights that Mythos Preview sometimes flat-out refuses to assist with AI safety Research & Development.

When asked to help researchers align models better, it actively pushed back.

The era of simple chatbots is officially over.

[l1t0 @dcanerate]

@heymike777 biggest miss is the shot you've never taken

[l1t0 @dcanerate]

@mattytay Incredibly smart idea

[l1t0 @dcanerate]

@pchopra28 where do i send de..?
nvm, you already have it

[l1t0 @dcanerate]

@pchopra28 that's good!

thank you!

[l1t0 @dcanerate]

@fdotinc Are you crypto-friendly? 👀

[l1t0 @dcanerate]

@fleetwood___ so accurate, amazing

[l1t0 @dcanerate]

@vibhu building a solid, future-proof backbone for the agentic economy on Solana must not be limited solely to providing more APIs.
we need to rethink the CORE principles of how data is designed in the first place.
AGI won’t be satisfied with 10 API endpoints or a raw datastream (IMHO)

[l1t0 @dcanerate]

@boardyai be a stanford grad

[l1t0 @dcanerate]

@annanay selling alpha to hedge funds? pff.
selling alpha OpenClaw Karpathy AutoResearch found on Polymarket for 10$ in compute to X audience? that's the real deal.

[l1t0 @dcanerate]

@mango_ good insights, my friend.

what's even less people understands is that after grants, community, other hackhaton, comes fundraise.

and fundraise is still 5% of progress towards final goal.

for whats in other 95% we should ask whoever was there :D

gl on your next venture!

[Boring_Business @BoringBiz_]

One of the greatest pieces of media ever created