ID

+1%

[ID @Djureth]

@rei_labs Grei

[ID @Djureth]

Well said.

"$REI/@rei_labs
is one of the most technically advanced AI agent projects in the entire space."

[0xHades @OnlyHades_]

Just a matter of time until people realize that $REI / @rei_labs is one of the most technically advanced AI agent projects in the entire space.

Core, their reasoning system, is built from multiple parallel components that operate across a dynamic and revisable knowledge structure, which means that it does not rely on a single model thinking in isolation. Instead, it distributes reasoning across a system that can update and reorganize what it knows over time.

The deeper thesis here is that $REI is focused on the actual mechanics behind intelligent agents:
reasoning,
memory,
instruction handling,
persistence,
and system coordination.

One of the biggest areas they’re tackling is the language boundary problem.

Human instructions naturally contain multiple layers at once: preferences, corrections, task rules, future expectations, scope limitations, contextual nuance.

REI is building mechanisms that separate and route each instruction to the correct internal layer with the appropriate level of persistence and behavioral impact.

Another important piece is the separation between reasoning and articulation.

Core produces standardized internal outputs independent of the language model sitting on top. This allows the reasoning layer itself to remain stable while the articulation layer can evolve independently.

The intentionality and technicality of this project is unmatched.

The 0.5 series updates (expected around the end of H1) will be a huge catalyst.

Once a working version of Core/reasoning architecture reaches more users, the market will reprice the token.

Less than 6 weeks until we see a god candle here.

[ID @Djureth]

@cryptic_tits Grei

[ID @Djureth]

@Gorev111 @rei_labs Grei

[ID @Djureth]

$REI is very aligned with Karpathy's idea of a cognitive core focused on reasoning rather than memorizing noisy data.

@rei_labs has been building the Rei Core for a long time. Inference-time learning is already showing strong results in real usage.

Team doesn't rest, Core 0.5a just dropped with per-unit evolution and much better recall/retention.

Grei

[Milk Road AI @MilkRoadAI]

Andrej Karpathy just made one of the most interesting arguments about AI model design that most people are completely missing.

His take is that frontier AI models are not too big because the technology is complex and too big because the training data is garbage.

When you or I think of the internet, we picture Wall Street Journal articles, Wikipedia entries, serious writing.

That is not what a pretraining dataset looks like.

When researchers at frontier labs look at random documents from the actual training corpus, it is stock ticker symbols, broken HTML, spam, gibberish.

One estimate puts Llama 3's information compression at just 0.07 bits per token meaning the model has only a hazy recollection of most of what it trained on.

So we build trillion parameter models not because we need a trillion parameter brain but because we need a trillion-parameter compression engine to squeeze some intelligence out of a firehose of noise.

Most of those parameters are doing memory work, not cognitive work.

Karpathy's prediction is separate the two entirely.

Build a cognitive core, a model that contains only the algorithms for reasoning and problem-solving, stripped of encyclopedic memorization and pair it with external memory that it can query when it needs facts.

He thinks a cognitive core trained on high-quality data could hit genuine intelligence at around one billion parameters.

For reference, today's flagship models run between 200 billion and 1.8 trillion parameters with most of that weight dedicated to remembering the internet's slop.

The trend is already moving his direction. GPT-4o operates at roughly 200 billion parameters and outperforms the original 1.8 trillion-parameter GPT-4.

Inference costs for GPT-3.5-level performance dropped 280-fold between 2022 and 2024 driven almost entirely by smaller, cleaner, better-architected models.

The real bottleneck in AI right now is not compute but rather data quality.

[ID @Djureth]

@MilkRoadAI @rei_labs

[ID @Djureth]

@rei_labs Grei

[ID @Djureth]

@elonmusk What until they discover Cyrillic.

[ID @Djureth]

@Lojom53 Grei

[ID @Djureth]

$REI keeps delivering results.
Once everyone gets access and more people start using it for all kinds of things, we're gonna see some really cool stuff.

[Rahul @RHLSTHRM]

Probably the most advanced AI trading agent system I've come across. I had the pleasure of helping @A_Keyboard cook this up in the early days, and it's advanced significantly since then.

@rei_labs is doing some crazy things on the backend to build a system that works fundamentally different than a normal LLM based architecture. This is not a simple case of passing a prompt + some data to an LLM and making it trade, this is a truly advanced real-time learning system.

Give it a shot if you can, this is the fundamental layer of the future where we will all have bots that manage our portfolios.

[ID @Djureth]

@0xreitern @rei_labs Looking good

[ID @Djureth]

Grei read if you're new to $REI

[Guidø @guid_o]

https://t.co/aqVB7oJidp

[ID @Djureth]

Interesting fact.

Tesla was a Serb.

His father, Milutin Tesla, was a priest in the Serbian Orthodox Church, and his mother Đuka (Georgina) Mandić came from a family with Orthodox priest heritage as well.

The three-finger gesture (often called the Serbian three-finger salute) is a widespread national/ethnic symbol for Serbs, commonly seen at sports events, celebrations, protests, political rallies, or any large gathering.

In religion, the sign of the cross is made using three fingers together to symbolize the Holy Trinity.

The number three repeat frequently in Serbian customs and folklore, reflecting its religious significance.

People greet close friends/family with three kisses on the cheek (alternating sides).

Certain superstitions or rituals involve actions in threes (e.g., spitting three times to ward off bad luck). etc.

[ID @Djureth]

@levelsio See @rei_labs

[ID @Djureth]

GRei

[Rei @rei_labs]

https://t.co/z7NITOUujD

[ID @Djureth]

@rei_labs Grei

[reitern @0xreitern]

Core's NeuroEvo simplified

Self-optimizing AI infrastructure that perpetually evolves as concurrent users shape it in real-time

[Realis Worlds @realisworlds]

People don't understand the scale of $RLS

63,000+ km² - larger than most game worlds combined

Built from real Earth data. Not procedural generation

This is the canvas for AI civilization

[Rei @rei_labs]

There's growing momentum around "data poisoning", deliberately corrupting the datasets that models learn from. The logic is straightforward: intelligence lives in the weights, so corrupt the weights, corrupt the intelligence. The model absorbs everything indiscriminately, unable to distinguish signal from sabotage, and reproduces the corruption indefinitely. The attack surface is the entire internet.

This vulnerability exists because current AI is fundamentally regurgitation at scale. Trillions of tokens compressed into statistical pattern matching. The model never "understands”, it memorizes and interpolates. There's no deeper cognitive process to catch this type of corruption.

Core is architecturally resistant on two fronts.

Units are blank slates. They carry no pre-baked knowledge scraped indiscriminately from the web. Intelligence emerges through reasoning over data inferred directly from you, your context, your inputs, your domain. The poisoned well everyone drinks from simply isn't part of the architecture. You bring your own water.

Core's center is an inference engine, not a knowledge repository. What evolves are strategies, reasoning methods, approaches to connecting ideas. Successful patterns survive and propagate. Failed patterns are eliminated. Selection pressure operates continuously. A dataset can be poisoned because it's static, it sits there, inert, trusted by default. A strategy can't be poisoned the same way because it's subject to ongoing evolutionary pressure. Bad reasoning paths lose. They don't reproduce.

More on NeuroEvo Below

[ID @Djureth]

GREI

[Plata @platacrypto]

https://t.co/OYqWTM1DVf