Anand Kulkarni

Built Gita Flow app for iOS | AI learner, critic, evangelist

[Anand Kulkarni @iamabk]

GPT 5.6 Sol new pricing.

Sol now costs $4 per million input tokens and $20 per million output tokens, 20% lower input pricing and 33% lower output pricing.

[Anand Kulkarni @iamabk]

@svpino I migrated from Codex to Claude Code (Max 20x) couple of weeks ago and realising that it wasn’t a very good switch.

[Anand Kulkarni @iamabk]

Irrespective of what the Darios and Sams of the world are saying, AI is nowhere close to make ANY job redundant yet.

Anybody disagrees?

[Anand Kulkarni @iamabk]

Agree.

Sol digs a nice rabbit hole and drags you along for long before you even realize.

Fable is flash in the pan…works as long it works.

Opus 5 - wonder what it is for.

Opus 4.8 new version: used and wouldn’t even touch again

I have all my hopes on Gemini now. I am actually happy that they are staying away from the drop-something-now rat race…hopefully they are working heads down on something that is actually good at coding 🤞

Wonder if Meta can come strong in the coding space.

What do you think?

[Anand Kulkarni @iamabk]

Sir Dario, please do not worry about rare diseases yet. Your Fabel and Opus are causing enough stress and agony to coders. Please cure that first.

I am sure you will earn a place in history for curing cancer but Anthropic is nowhere close to that yet. A step at a time staring with coding please….

Respectfully

[Anand Kulkarni @iamabk]

@DarioAmodei

[Derya Unutmaz, MD @DeryaTR_]

Mr. Amodei, I appreciate that you care so much about us that you even came to X to write this long post, despite how much you seem to hate being here among us. 🙏

But Mr. Amodei, please don’t worry at all. We, the rest of humanity outside you and your company, are fully capable of curing cancer and every other disease using AI, defending ourselves against malicious actors misusing AI, and pushing humanity toward a golden age of abundance, even if you and Anthropic were to suddenly disappear tomorrow.

Thanks to great American and Chinese AI companies, extraordinary engineers and scientists from around the world, and millions of people harnessing open-source models, we really can figure out what is best for humanity and push ourselves toward the best timeline in human history. You need to believe in us, trust us that we will do the right thing!

We simply ask that you not interfere with that progress or advocate for regulations that restrict our freedom to use the best intelligence available that we can afford. Important to note that, when such efforts are made to restrict that freedom, we will push back as strongly as possible to protect our right to access the best intelligence at the lowest possible cost, so that every human being can be empowered by it.

You may, however, want to focus on your company: rightly so, creating enormous wealth for your investors and employees, making sure your models remain continuously at the forefront of AI, and helping advance AI security, accessible by all and not just selected few or those who can afford it, in other words in ways that create maximum value both for yourself and humanity.

Thank you for your attention to this matter.

[Anand Kulkarni @iamabk]

Genuine question: AI saw the adoption because it improved productivity and saved time. Now, if you save time, you will be penalised. How does this make sense?

Why does AI text, image, video generation even exist if users are not supposed to use it anywhere?

[Thariq @trq212]

This is part of working with the EU AI Act, other labs are adding similar watermarking.

It’s hard to identify AI-generated text, and this gives people better tools to do that. We’ll also be a shipping text detection API that you can use yourself.

[Anand Kulkarni @iamabk]

Experiencing lots of behavioural (and probably architectural) similarities between Sol and Opus 5.

Sheer coincidence??? 🤔 @AnthropicAI @OpenAI

[Anand Kulkarni @iamabk]

@vishalsingh2972

[Anand Kulkarni @iamabk]

Forward Deployed Engineers: hey guys, we’ve built powerful AI solution. We are now on a mission to find problems it can solve

[Anand Kulkarni @iamabk]

Despite the deafening (paid) hype, coding is still an unsolved problem.

FYI please @finkd @miramurati

[Anand Kulkarni @iamabk]

In a race to launch newer models, both OpenAI and Anthropic seem to be dumping untested models, which are not even fully thought through.

Sol 5.6 and Opus 5 are examples of that.

I respectfully urge both to take their time for progressive lunches, not regressive.

[Anand Kulkarni @iamabk]

@sama This is great, Sam. Thanks. But Sol seems to be happy digging its own rabbit holes and enjoying it.

[Anand Kulkarni @iamabk]

Gemini’s long silence is actually giving me hopes…that the team is working heads down on something progressive.

Google being Google, you can never be sure of their short-lived experiments. Remember 3.1, nano banana overhype?

[Anand Kulkarni @iamabk]

In a race to launch newer models, both OpenAI and Anthropic seem to be dumping untested models, which are not even fully thought through.

Sol 5.6 and Opus 5 are examples of that.

I respectfully urge both to take their time for progressive lunches, not regressive.

[Anand Kulkarni @iamabk]

@svpino I did that, used Sol xHigh for a few days and now back to Claude.

[Anand Kulkarni @iamabk]

@RaoulGMI @AnthropicAI Opus seems to have learnt it from Sol.

[Skyroot Aerospace @SkyrootA]

ORBIT ACHIEVED. 🚀

Vikram-1 Test Flight-1 has reached orbit. India's first privately developed orbital rocket has completed its final burn and injected its payloads into a ~450 km orbit, making India the third country in the world with private orbital launch capability.

History is made. 🇮🇳

#Vikram1 #JourneyToOrbit #SkyrootAerospace

[Edward Lando @edwardlando]

The startup game is both beautiful and cruel.

Hung out with a founder friend over the weekend who's worked 70-hour weeks for the last 6 years, raised tens of millions from tier one investors, hired a very strong team, and has been building in what is generally a pretty good space (b2b software).

Just never really found full PMF despite pushing a boulder up a hill to millions in revenue.

He seemed exhausted and somewhat defeated, and this is someone who embodies the literal opposite of these qualities.

I can tell that he still knows he will do something great in the long run, because he knows he is capable of it, and wants to prove it to himself.

But at this present moment he is very sapped of energy.

You read a lot about the fast unicorns (and decacorns) but he is only one of tens of thousands of redoubtable founders who grind on something for a long time and don't quite make it.

This is the dark matter of the startup universe (as he calls it, actually.)

It’s simultaneously omnipresent and invisible.

People only ever talk about and project the “we’re crushing it!” stories.

But almost every successful entrepreneur has a story like this one to tell, where they spent years grinding in relative darkness.

Excellence and luck: you need a lot of both for something to work.

If you play the game long enough, and if you never lose hope, you probably will get both at some point, and make it to tell your own success story.

[Anand Kulkarni @iamabk]

@anandlunia In my view Claude, Gemini (Google AI Studio), OpenAI are already building their vibe coding tools and they will dominate going forward. Replit is good and is making very thoughtful strategic moves - recently collaborated with Accenture and Microsoft.