Pankaj Agarwal (d/acc)
749 posts

Pankaj Agarwal (d/acc)
@Shankusu993
Building @CortexeaAI | Always learning, tinkering and building | CSE grad @IITKGP
on chain Katılım Aralık 2018
908 Takip Edilen309 Takipçiler

@OpenAI On a followup note, it's priced cheaper but tend to use too many reasoning tokens (and seem to require higher token limits to work well) adding up to a higher cost. Interesting, is this a new age dark pattern that LLM companies are coming up with?
English

While @OpenAI gpt 5 didn't meet the high expectations set prior to launch, my experience using it for the past few days has been pretty good. CFG support is a big plus (still evaluating how well it works, but it has many implications and I like the direction taken here)
English
Pankaj Agarwal (d/acc) retweetledi

Sentient’s mission is to ensure that Artificial General Intelligence is open-source and not controlled by any single entity.
We’re launching the GRID: the world’s largest network of intelligence.
The GRID makes open-source the default for AI development and ensures humanity wins the future of AI.
Let’s dive in👇
English
Pankaj Agarwal (d/acc) retweetledi
Pankaj Agarwal (d/acc) retweetledi

Happy 10 @ethereum ! 🎉🎊
Celebratory moment for entire Blockchain/crypto community as the world's first decentralized programmable ledger turns 10 with a 0 downtime history! Quiet crazy what the journey has been, and only more to come!
#ethereumturns10
English

@prashant_xyz Much respect ser 🫡 thought I only witnessed a short part of spheron's growth closely, you guys always stood out as real builders creating value with much dedication and relentless hardwork.
Many congratulations on the upcoming milestone! 🎉
English

TGE Reflection — Thank You to the Builders Who Shaped Us
As we approach TGE, I wanted to take a moment to reflect — and more importantly, say thank you.
Every ecosystem we touched, every relationship (good or bad), every challenge — helped shape what Spheron has become today.
My journey in crypto started with @ArweaveEco . @samecwilliams was the first person who made me believe in the potential of decentralized infra. The entire Arweave team was phenomenal — they helped us form our earliest mental models.
We couldn’t build a long-term vision there, but we did contribute one thing: 1-click website deployment on Arweave.
We then moved to @Filecoin , where the learning deepened. @juanbenet once told us, “You’re building something special — and Filecoin will back you.” That one call gave us immense clarity.
The @protocollabs @TachyonXYZ & Filecoin fam stood by us even when we weren’t sure of our direction. That support meant a lot.
Next came @akashnet — and I’ve always admired what @gregosuri and the team built.
We built on Akash. We studied the code. We scaled on it. We even pitched systems like the "battery mechanism" to the team — and they genuinely appreciated it.
There was real collaboration. There was real value added.
Unfortunately, that chapter ended on a sour note.
Maybe it was a difference in vision. Maybe just timing.
Either way, I want to say this: if anything from my side contributed to that friction — I apologize.
We learned a lot from Akash. It inspired us to go deeper. And that’s something I’ll always respect.
We’ve since built our own architecture, published an open-source whitepaper, and created a full-stack system — independent and community-first.
But none of this would’ve been possible without the early lessons, feedback, and relationships formed in these ecosystems.
With TGE around the corner, it feels like we’re handing over a grown-up version of Spheron to the community.
And I won’t lie — it makes me nervous.
But it also makes me proud.
Because now, this protocol will live, grow, and evolve with the community at the center.
Post-TGE, Spheron Day 1 begins — again.
I'll be working even harder to bring demand to decentralized compute. Not just on @spheron , but across @akashnet , @ionet , @AethirCloud , and beyond.
This is a shared mission — and we all have a role in making decentralized infra win.
Thank you for helping us get here. 🫶
English

@jito_sol Quiet a few changes/improvements, private txns, verifiable ordering and plugins are super cool
English

@Saxenasaheb True, deeper and on-ground insights would come from talking to these game devs, that's just my hypothesis and initial approach.
I think the main problem here is for gamers- but if we think for game devs it solves capital fragmentation and (some) new demand for early stage
English

@Shankusu993 Tough yaar. Once you take with these games you’ll have deeper insights here.
Worth jamming with a few games esp early stage to see if this is a challenge for them.
English

@Saxenasaheb This hypothesis and overall system flow will still be valid, and can be tested with any game, so these indie games + any game with open payments api, or on-chain games/payments (need to hunt these).
Also, OP promised me the assumption of users 😅, means the game dev get them too
English

@Saxenasaheb It would be great if there was someone building in my existing network, but if not my best bet would be indie game dev forums.
They have clear incentives - get exposure to more users, and get direct access to gamer's funds without a credit lock-in (ease of tnx).
English

@Saxenasaheb (2/2)
To test the core hypothesis of cross platform and convenience, these are also must:
- at least 2-3 games integration/support, ideally both on-chain games but also the traditional ones
- 1 click payment support on the payment interface
- Standard security practices
English

@Saxenasaheb (1/2) I beleive, these are a must have from an MVP standpoint:
- A wallet interface to load funds and track payments for gamers
- Stable coin support, 1 popular one should suffice
- A payment gateway API/SDK for integration with partner platforms
English

@alexanderlong That makes sense, but I also feel it's similar to any startup story. Whoever survives longer wins, china doesn't have to remain open forever, just a few years longer than west, once projects have moved to their models, the friction of moving back to US models will be v-high.
English

@Shankusu993 Thing is i think the probability of Chinese models continuing to be openly released long term is also very low.
English

Feel like meta closing models was very predictable. I explicitly said this would happen last year and explained why (from blog.pluralis.ai/p/article-2-pr…).

Shane Gu@shaneguML
RIP to the unicorn AI startups that have zero products, zero foundation models, and were just going to depend on big labs releasing open-source models for free to model merge. I know one or two.
English

@eigencloud Roses are red,
Violets are blue,
I use EigenCloud,
You should too
English


