@radoshi

ChatGPT Finances @chatgptapp Formerly: @hirofinanceai (acq @openai) @hellodigit @facebook @google @microsoft Writing: https://nitter.cf/t.co/VWRK2EDon9

San Francisco
Joined February 2008
Rushabh Doshi retweeted
The Next Rung At some point I realized that almost all of the code I had ever written had been deleted. At the very least, refactored or rewritten, but my LOC were for the most part confined to the dustbin of repo history. I was lamenting the fact that the stuff that did survive felt like it was all some embarrassing junk I had thrown together in a couple of days before moving on to the next thing, and I wondered what mark I had really left on the code. The trend accelerated once I became a manager and my dwindling contributions were swiftly consumed by normal code churn. Most of the coding I did was nights-and-weekends tinkering on personal projects. I was turning all this over in my head and spiraling a bit thinking about what really makes a great software engineer and was I ever one and would I ever be one. I realized the people I admired most weren't the galaxy-brained folks who built magnificent "future-proof" systems that would still work in ten years but the ones who could see just one move ahead. When the next big problem arrived, their design was just malleable enough to make the transition straightforward. Their code felt almost designed to churn. Great software engineering is building rungs on a ladder. The code that got mowed over in a refactor had to exist for the refactor to happen. It served its purpose and now sails off in a Viking funeral. What is the alternative? Climbing back down? All that code I wrote to indulge myself on nights and weekends? I still have absolute git blame dominion over it. 100% of the lines have my name on them. And they do not matter at all, and the way you can tell is that they're not churning. Many of those repos haven't seen a commit in a decade. With vanishingly few exceptions, churn is a sign of life. The stuff that doesn't last is the stuff that matters. — "The models are the worst they will ever be." I hear from my friends at the labs that some researchers act almost as if programming (even with agents!) is a waste of time, because the models are going to eventually just be able to do it all for you. Yes, of course nobody writes the code by hand anymore. Some people have given up on reading it altogether. Some people don't even try to understand how it works anymore. But it is a mistake to assume that just because the code is disposable, the understanding is too. For my part, the amount of handwritten code I'm producing does indeed round to zero. But not understanding what's actually going on underneath is insanity for anything non-trivial. You might be able to get away with this someday, but you cannot build a system of any complexity whatsoever using even the best models without good technical guidance, and you cannot give good technical guidance if you have no idea WTF is going on. Maybe we will get to the point where you can just let go and let Claude. But we are not there yet, and we will never get there if we try to operate as if we are already living in some imagined future instead of the real one we live in today. There is a difference between being able to see a rung and being able to grab it. We have to keep building. We have to keep climbing. fic.us/writing/the-next-rung
1
1
1
12
1,071
Rushabh Doshi retweeted
ChatGPT Finances now supports @Coinbase connections through @Plaid! This has been one of our most requested features. Thank you for your patience.
20
12
6
150
17,874
Health and Finances Go make your life better.
Helpful (and dare I say fun?) new feature in ChatGPT Health: tap on anything in your synced data and get a helpful summary of everything related to that metric. No prompt required! Makes use of your chat-wide memory, so can be incredibly personalized and specific to you.
12
643
You’ll want to listen to this.
My next podcast episode is almost guaranteed to save you some money. It's with my friend @ebloch, product lead for ChatGPT Finances at @OpenAI. I asked him to share six practical personal finance prompts that anyone can try to: 1. Find tax savings proactively 2. Maximize credit card rewards 3. Create interactive forecasts like your path to financial independence ...and more 📌 Subscribe to get our full episode tomorrow: youtube.com/@PeterYangYT?sub…
1
1
4
1,084
Got some stocks you’re tracking? Add them to the watchlist, get daily summaries of movements.
Stock watchlist rolling out in ChatGPT + Finances for US Plus/Pro! Tell Chat what stocks to follow. It’ll keep an eye on what’s happening and proactively bring you updates. Try: “@Finances add Nvidia, Microsoft and Berkshire to my watchlist.”
1
2
344
Rushabh Doshi retweeted
Verified U.S. clinicians can get free GPT-6 Astra (Pro) access today via ChatGPT for Clinicians. Sign up: chatgpt.com/plans/clinicians…
204
258
140
2,748
1,247,645
Rushabh Doshi retweeted
we used our models to find and fix critical vulnerabilities in our own systems, as part of a 250+ person effort within openai. playbooks/learnings/architecture which we hope can be helpful for others to make the most of the defenders' window:
We mobilized 250+ people to strengthen our defenses across hundreds of systems. Our latest cyber models helped us find and fix vulnerabilities we might never have discovered otherwise. We’re sharing what we learned, the architecture, and a practical playbook so you can build your own Defense Factory: a continuous loop where AI agents find vulnerabilities, validate them, and verify that fixes work. openai.com/the-defense-facto…
111
94
9
1,456
143,668
Ask @ChatGPT finances how much money you need in your checking account and sweep the rest into a money market or account earning account> 3%. I just paid for your annual ChatGPT for about 5 years.
2
7
595
I’ve done the same.
ChatGPT is getting kinda insane now. Was buying tires at Costco yesterday and they tried to upsell me to Executive. So I just asked ChatGPT if I should. It looked at my actual spend: $4,751 last year, $3,029 YTD. Then calculated the exact breakeven: $3,250/year, and pointed out I was already well above the breakeven pace this year, before even buying the tires. Pretty useful.
3
989
Rushabh Doshi retweeted
GPT-6 Astra is our healthiest model yet. ♥️ It sets a new state of the art on HealthBench Professional, our benchmark of real clinician tasks, including care consultations, medical documentation, and research. At its lowest reasoning effort, Astra already surpasses GPT-5.6 Sol’s best score at about half the cost. This evaluation is intentionally designed to be challenging for our models and focuses on especially complex and rare situations. On another internal evaluation more representative of health questions people often ask ChatGPT, Astra was more than three times less likely to make factual mistakes than GPT-5.6 Sol at their strongest reasoning settings. We hope this release is another step towards trustworthy, abundant health intelligence for patients and clinicians.
33
36
8
399
112,504
Rushabh Doshi retweeted
GPT-6 Astra is here. We hope it will begin to enable a new generation of entrepreneurship, scientific discovery, and building. We believe it is the best model in the world for computer use, professional work, science, coding, cybersecurity, and more. It took us some extra time to ensure that we could meet the safety and alignment standards required for this capability level, but we think you’ll find it worth the wait. It scores 98% on FrontierMath Tier 4, 99.9% on ARC-AGI 3, and 100% on ExploitBench.
2,261
4,406
1,198
55,528
4,810,946
Luna is an incredible model
Everyone's talking about Astra, but Luna is over here crushing it with very little fanfare – #1 on OpenRouter! It's a good good model.
1
339
Rushabh Doshi retweeted
The evals are pretty wild, but there's a more visceral feeling you get when you see Astra do computer use for the first time. That was the piece that was most mind-blowing for me. Once you get a chance, would recommend trying computer use in the ChatGPT desktop app with Astra.
Replying to @OpenAI
GPT-6 Astra is state-of-the-art on FrontierMath Tier 4, ARC-AGI 3, and TerminalBench-4.0. GPT‑6 Astra is also a major advance for scientific discovery, with state-of-the-art performance on Terminal-Bench Science 0.1 and HealthBench Pro.
17
6
1
187
8,320
Rushabh Doshi retweeted
GPT-6 Astra represents a step-function change in model capability for interactive reasoning problems. It scores 66% on ARC-AGI-3 using our standard harness, and nearly 100% with a continuous conversation harness and custom compaction, at a cost of roughly $360 per game. In fact, the continuous harness version significantly outperforms our human baseline in action efficiency across almost all levels. When we examined the reasoning chains to understand how the model operates, we found it performing highly efficient, on-the-fly symbolic world modeling for each game and level. It goes as far as developing its own shorthand DSL to represent in-game situations -- essentially a game-specific algebraic notation. Overall, Astra exhibits symbolic modeling behaviors we had previously only seen with sophisticated harnesses -- so harness capabilities are increasingly shifting into the model itself. We see Astra as a major breakthrough in model intelligence. Read our post on Astra and what these results mean: arcprize.org/blog/astra
195
871
258
6,952
983,034
This guy …
who wants one? share your favorite ChatGPT + Finances use case and I’ll send you one of our precious stickers.
2
5
689
This is the way
Replying to @jordancooper
I finally hooked up ChatGPT to Apple Health and it has been 100x more useful
1
1
372
🚢🚢🚢
54 fixes and improvements to ChatGPT Finances this past week 🔥 A few highlights: • faster transaction detail loading • more accurate spending totals • category corrections that carry through consistently And yes, we hear you on desktop and international expansion. What's the one thing you most want us to fix next?
10
817
Cannot stress how important it is to take this seriously. OpenAI, Anthropic, Google, AWS, Microsoft all coming together.
An open letter for a global surge in cyber defense, signed by over 100 organizations including Anthropic, AWS, Google, Microsoft, OpenAI, and Oracle.
2
6
651
Don’t forget to reconnect with the hundreds you’ve lost touch with
The most underrated way to make new friends at any age is to just do the same stuff at the same times at the same places on a regular basis
2
191
Got one!
We built a small biped robot you can teach new tricks to. Train it in simulation, run it on the real thing. Meet Microduck 🦆 $399, shipping before Christmas. pollen-robotics.com/microduc… github.com/pollen-robotics/m…
1
5
643