it costs $6000 a month to run a single 24/7 OpenAI realtime voice discord bot
think about that
william
2,823 posts
william
@wgussml
Project Prometheus, prev cofounder @generalagents, OpenAI, helped build copilot & MineRL, creator of ell
- It's here!!🎉🚀🎉 Say hello to GhostWrite - Your ChatGPT powered AI email assistant 👻 GhostWrite automates your inbox w/ AI replies, email, and summaries. Write emails 20x faster & get to inbox 0️⃣ w/ GW! 🔗chrome.google.com/webstore/detai… Retweet & comment below for unlimited access!
00:00 - tech debt deflation is here o1 pro just solved an incredibly complicated/painful rewrite of a file that no other model has ever gotten close to I’ve been using this an eval for different frontier models & this marks a huge shift for me we’ve entered the “why fix your code
- 🚀 I'm excited to announce the future of prompt engineering: 𝚎𝚕𝚕. developed from ideas during my time at OpenAI, 𝚎𝚕𝚕 is light, functional lm programming library: - automatic versioning & tracing - rich local oss visualization tools - multimodality native Read on ⬇️
00:00 - y'all don't know but the xAI team literally doesn't sleep this is no joke cracked.Elon Musk claims to have finished a 100,000-strong H100 cluster in four months. How likely is that?
- tomorrow is going to change everything
- excited to announce that pro users will get 1 request for o3 pro per week! huge kudos to the inference teams that worked tirelessly to make this happen
- we are building the most incredible model right now it is magical
- this is kinda crazy but I think I’m more excited for Gemini 2.5 flash than I am for gpt 5
- 🚀 The first realtime AI that pilots your desktop is here: Ace Excited to share what I've been helping build at @GeneralAgentsCo over the past few months Read on ⬇️
00:00 - o3 is in insane i dont even know what to say this is the moment ive been waiting for in programming really hard to comprehend what o4 will feel like
- what most people will miss is that o1 is significant precisely because it isn’t an SFT on synthetic data the fact that rl on CoT unconstrained works and doesn’t collapse to gibberish cot steps is really profound more thoughts ⬇️Claude Sonnet 3.5 can do Strawberry with the right prompting. Guys, there's no moat. There's no secret sauce. We can synthesize the data with any model.
- o1 is so incredible for learning pure math I am in awe my perspective on oai utility and current advantage is completely changed, yes others will catch up soon but wow this model is different
