Klein Field

@chatgpt21 Free return until Jan 31 too

Walmart selling unitree robots is just the start. Soon there will be many highly intelligent bi-pedal humanoids available for sale that aren’t just teloperated remote control clankers.

@MatthewBerman What games are you playing on your switch?

I'm ready for my 2.5 hour flight: ✅Macbook Air ✅iPad Pro ✅Remarkable Pro ✅Airpods 3 Pro ✅iPhone 17 Pro Max ✅Switch 2

@aidan_mclau @apples_jimmy @tszzl @sama Had the context window for Canvas been improved at all? Would be better for those longer outputs + code gen edits

@apples_jimmy @tszzl @sama output length is actually quite improved, you can request and get longer outputs than any previous openai model

we've been testing some new methods for improving writing quality. you may have seen @sama's demo in late march; GPT-5-thinking uses similar ideas it doesn't make a lot of sense to talk about better writing or worse writing and not really worth the debate. i think the model writing is interesting, novel, highly controllable relative to what i've seen before, and is a pretty neat tool for people to do some interactive fiction, to use as a beta reader, and for collaborating on all kinds of projects. the effect is most dramatic if you open a new 5-thinking chat and try any sort of writing request for quite some time i've wanted to let people feel the agi magic I felt playing with GPT-3 the weekend i got access in 2020, when i let that raw, chaotic base model auto-complete various movie scripts and oddball stories my friends and I had written for ~48 hours straight. it felt like it was reading my mind, understood way too much about me, mirrored our humor alarmingly well. it was uncomfortable, and it was art base model creativity is quite unwieldy to control and ultimately only tiny percents of even ai enthusiasts will ever try it (same w the backrooms jailbreaking that some of you love). the dream since the instruct days has been having a finetuned model that retains the top-end of creative capabilities while still easily steerable all reasoning models to date seem to tell when they're being asked a hard math or code question and will think for quite some time, and otherwise spit out an answer immediately, which is annoying and reflects the fact that they're not taking the qualitative requests seriously enough. i think this is our first model that really shows promise at not doing that and may think for quite some time on a writing request it is overcooked in certain ways (post training is quite difficult) but i think you'll still like it 😇

@bobby_io @OpenAI But what is the context window in canvas? Probably the limiting factor

ChatGPT canvas with the new GPT-5 model looks a lot like Lovable, Bolt, Replit, v0... But also with: - conversational voice ai - deep research - Codex - All your past chats, memories - projects etc, all in 1 place Yo @OpenAI can you add a import from github button? Lets cook!

@ahmetbuilds What about in Canvas? Is the limit the same; Like 300-400 lines of code/ how many tokens can you paste into it?

GPT-5 context windows in ChatGPT: 8k for free users, 32k for Plus, 128k for Pro same numbers as it was before with gpt-4o, bummer they didn't increase it since the model now has 400k context window

@kimmonismus Is the context window the same in canvas? Like how much context can you paste into canvas before it says it’s too much?

It’s live for me - in Germany!

@unusual_whales Ok but what for? How do we get the energy to the earth? Or is it for use in the moon?

US has announced plans to build a nuclear reactor on the moon.

@testingcatalog Ask the black hole question that Xai showed on stage yesterday

Which prompts should I try? 👀👀👀

@chatgpt21 Maybe the moat is now the people behind the product and the way they make you feel, rather than only the product itself. It sort of makes marketing more important than the function/product itself

> Cluey raises 15M > marketing machine > constant hype > random guy on Reddit open sources it for free > this is the world with open source ai now

@sama Not my prompts lol ya boi using a full teaspoon 💦

also, here is one part that people not interested in the rest of the post might still be interested in:

wrote a new post, the gentle singularity. realized it may be the last one like this i write with no AI help at all. (proud to have written "From a relativistic perspective, the singularity happens bit by bit, and the merge happens slowly" the old-fashioned way)

@elonmusk @danawhite

@joshcirre You could probably make your money back by getting the $200 Gemini No wonder you family is dying You gotta spend money to make money

someone help me budget this my family is dying rent $1,200 groceries $400 utilities $150 Claude Max $200 car payment $350 ChatGPT Pro $200 phone bill $80 Cursor $20 insurance $180 Perplexity Pro $20 gas $120 Gemini Advanced $20 gym membership $40 GitHub Copilot $19 Netflix $15 Cursor Overage Credits $138.47 Midjourney Pro $120 coffee $60 v0 Pro $20 internet $70 Replit Core $20 groceries (2nd trip, forgot precut watermelon) $100

@_catwu When will Claude Code get a better conversation picker? Had to script one myself to browse/summarize old chats. First repo ironically couldn’t have been built without #claudecode 🤖 github.com/espositov/clau…

Highly recommend #claudecode for system file management. Have it access a folder, and rename files based on the files content. Very useful for random screenshots or unorganized downloads folder.

Claude users with a Pro plan: You can now try Claude Code!

Not today, but maybe @sama will grace us tomorrow

@sama @chatgpt21 Better stop eating and start shipping @chatgpt21

@chatgpt21 just a little longer! worth the wait imo. if you send us your silicon rocks maybe it will be better.

Day 47. Food rations are getting low. I’m surviving off silicone rocks and the hums of the patience cave

Sooon

@AnthropicAI Has this actually been rolled out to anyone?

We're rolling out voice mode in beta on mobile. Try starting a voice conversation and asking Claude to summarize your calendar or search your docs.