Gabriel

CTO @routemesh | 🤖⚡️

[Gabriel @gtupak]

🚀 After more than an year of work, I'm happy to announce that we've officially launched our RPC Aggregator service!

You can now access all RPC providers using a single service at the best price on the market.

Using RouteMesh, you can sleep easy knowing automatic failover is done for you, along with node sync and data quality monitoring.

Try it today, there's no lock-in!

[RouteMesh @routemesh]

1/ Introducing RouteMesh, the RPC provider that doesn’t run any nodes but has access to the largest inventory of nodes globally.

Because our model is permission-less (providers don’t need to sign up explicitly), we offer a service that has:

Sub 10ms latency, support for 1000+ chains, usage based pricing, high rate limits and failovers baked-in.

Currently serving billions of requests per month👇

[Gabriel @gtupak]

@tangouninformed @0xSero I'm running the full unquantized model and still having issues. Speed is great (runs on an RTX6000) but I still don't find it reliable for day to day work

[Gabriel @gtupak]

@shawshank_rat @0xSero and your tool calling works and all?

[Gabriel @gtupak]

status: being the human in the loop

[0xSero @0xSero]

In 12 hours we've had 323 people sign our petition to protect local AI.

Open Source must win, not because anyone else must lose. But so that we can all win.

Please help me get 10,000 signatures so that we can walk into the room and say people care.
https://t.co/Z6WpNfSMXr

[Gabriel @gtupak]

@KyeGomezB how do you define AGI?

[Gabriel @gtupak]

impressive - good to remember when someone mentions that AI consumes a lot of water

[Chubby♨️ @kimmonismus]

A quick fact-check on AI’s water usage, including data centers, based on the cited source.

[0xSero @0xSero]

If you have the NVMe

Go download as many models as you think you might ever want. Now, go on Huggingface.

They’re coming for open models next.

[Gabriel @gtupak]

@jerallaire What are they buying?

[Gabriel @gtupak]

@zanoga @frankb737459085 What projects do you have? Haha

[Gabriel @gtupak]

@maczzzzzzzzzzzz @NousResearch It's respectable

[Gabriel @gtupak]

@aerugoettinea Give the diet in the 4 hour body by @tferriss a shot. I found it to be super effective

[Gabriel @gtupak]

@SakanaAILabs Has anyone ran a test doing a comparison of how may tokens it consumes compared to a task you'd normally give to opus for example?

[Gabriel @gtupak]

@robinebers By review workflow you mean a PR review? How big was your PR? I was able to do 19 decently sized PRs with the same amount of money (~$44), including addressing reviews made by CodeRabbit. Something seems off here...

[Gabriel @gtupak]

@grok @NewestPapa @guohao_li @grok what are the privacy concerns if one decides to use their api vs one that's hosted somewhere else like @FireworksAI_HQ ?

[Gabriel @gtupak]

@NewestPapa @guohao_li @grok how does https://t.co/F14AChceib plan to become a profitable company given they make open weights models?

[Gabriel @gtupak]

@iMuffined @__tinygrad__ If you dont buy the 1MW container you ngmi 😂

[Gabriel @gtupak]

@jeffbarg OpenChamber with OpenCode and Fireworks

[Gabriel @gtupak]

from what I'm seeing with agentic AI, we're at the phase of AI loops and open source frontier models

[goodalexander @goodalexander]

The thing people are missing on GLM 5.2

Before it would have been kind of useless to rent H200s bc the open models weren’t good and codex subsidies so high no practical need to

But subsidies getting cut. And GLM 5.2 is good enough to want *way* more tokens bc looping doesn’t break

Not to mention god knows what data you’re going to lose now the companies are retaining your data more for “national security”.

Completely changes economics in structural semi permanent way and it’s worth noting each of these points are likely to accelerate

1) looping (massive token consumption with sub agents) will get better with model capability and base error rate bc compounded error
2) coding plan subsidies will get cut as pressure builds into IPOs to show margin improvement
3) national security concerns will increase causing larger data retention windows and more IP risk and perverse incentives (Anthropic building the software you use Claude code to build)

Providers like Vast are already seeing insane hockey stick acceleration that started in Feb and is probably going to hit a parabola

This creates a flywheel for discount and efficiency providers which are highly international by nature (Malaysia bought billions of GPus, Middle East)

It’s worth repeating: prior to glm 5.2 this was all kind of theory / LARP. Just bc the open source models got brutally outperformed by close source

I’m incredibly jacked up. Rising tide lifts all boats. GPUs. Chip Marketplaces. Chip financing. Crypto mining. The future is now