Log inSign up
prashant
386 posts
user avatar
prashant
@prashantmital
doing stuff with AI, startups and my 🦮. all takes personal and low-quality.
San Francisco, CA
Joined July 2015
3,020
Following
2,826
Followers

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • Pinned
    user avatar
    prashant
    @prashantmital
    Jun 25
    @justoutquan is a builder with exceptional taste and values. i’ve felt his intention to build AI that drives **genuinely** good behavior everyday since i first started using Tomo back in late 2025. very excited for everyone to experience the joy of @tomo in their lives 🥹
    user avatar
    Justin Quan
    Tomo
    @justoutquan
    Jun 25
    It's been 3 years since AI exploded, but nobody's life has improved outside of Silicon Valley. Tomo is for real humans, like my little brother Lucas and 10,000 other humans taking a chance on themselves every single day.
    00:00
    199
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    feeling frustrated (and a little guilty): there’s still way too much confusion about @OpenAI's Responses API. this is partly on us: we haven’t always been clear about why we built it, how to use it, and why it matters. here's my attempt at setting the record straight. 👇
    545K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    Replying to @prashantmital
    myth #3: model intelligence is the same regardless of whether you use completions or responses wrong again. responses was built for thinking models that call tools within their chain-of-thought (CoT). responses allows persisting the CoT between model invocations when calling
    o3/o4-mini Function Calling Guide
    o3/o4-mini Function Calling Guide
    From developers.openai.com
    209K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    Replying to @prashantmital
    in summary: - responses = completions ++ - works in stateless & ZDR contexts - built for thinking models - unlocks higher intelligence and maximizes cache utilization in agent loops if you’re still on chat completions, consider switching now -- you are likely leaving
    31K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    Replying to @prashantmital
    myth #1: it's not possible to do some things with responses nope. responses is a superset of completions. anything you can do with completions, you can do with responses – plus more. for example, you can manage the conversation state manually (like in completions) OR let the
    40K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    Replying to @prashantmital
    myth #2: responses always keeps state, so we can’t use it in strict Zero Data Retention (ZDR) setups wrong. you can run responses in a stateless way. just ask it to return encrypted reasoning items, and continue handling state client-side as shown above.
    41K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    Replying to @prashantmital
    one founder's explanation for why they hadn't switched over to responses yet: we want fine-grained control over our agentic processes – we handle context construction and prompt caching optimization ourselves the reality? this is all possible with responses!
    36K
  • user avatar
    prashant
    @prashantmital
    Sep 4, 2025
    a few founders I’ve spoken to recently were under the impression that they need to go via Azure for inferencing OpenAI models if they have a use case in a sensitive vertical. this is FALSE. we offer ZDR and Modified Abuse Monitoring on our platform, have done for a while. to
    help.openai.com
    How can I get a Business Associate Agreement (BAA) with OpenAI for the API Services? | OpenAI Help...
    Information about HIPAA compliance for healthcare companies
    4.4K
  • user avatar
    prashant
    @prashantmital
    Sep 5, 2025
    my team has a new LI home, please follow: linkedin.com/showcase/opena… bonus: 📸 of the wonderful folks you get to work with when building your startup on @OpenAI!
    944
  • user avatar
    prashant
    @prashantmital
    Jul 23, 2025
    📈The hardest thing about this one was all the side quests we couldn't go on! Reward engineering and hacking, hyperparameter optimization and task selection for RL could all probably do with Build Hours of their own 😅.
    user avatar
    OpenAI Developers
    OpenAI
    @OpenAIDevs
    Jul 23, 2025
    The fine-tuning dashboard makes it easy to visualize how model performance shifts across checkpoints. Watch @promptshant and @TSautory use it to dial in precision vs recall for a nuanced legal classification problem. Watch the full demo from Build Hours: webinar.openai.com/buildhours/rei….
    00:00
    468
  • user avatar
    prashant
    @prashantmital
    Aug 11, 2025
    Replying to @simonw
    if it’s any consolation, we struggle with this internally as well (we will fix it) also, you forgot codex-1 which is a custom trained o3 variant that powers ChatGPT Codex 😅
    651
  • user avatar
    prashant
    @prashantmital
    Apr 16, 2025
    Super excited about @OpenAIDevs open-sourcing Codex CLI. This is huge if you have been building custom pipelines for code-related tasks like large codebase migrations because you can just branch off of our reference implementation to accelerate your build! #OpenAI
    user avatar
    OpenAI Developers
    OpenAI
    @OpenAIDevs
    Apr 16, 2025
    Meet Codex CLI—an open-source local coding agent that turns natural language into working code. Tell Codex CLI what to build, fix, or explain, then watch it bring your ideas to life.
    00:00
    422
  • user avatar
    prashant
    @prashantmital
    Jul 16, 2025
    🚀 Join @TSautory and I on today’s edition of OpenAI Build Hour. This week’s topic: Reinforcement Fine-Tuning (RFT) 🔍 📅 July 16 | ⏰ 10 AM PT 👉 Sign up: webinar.openai.com/buildhours RFT is great for teams working in law, healthcare, finance, or any domain where reasoning matters.
    OpenAI Build Hours
    From webinar.openai.com
    306
  • user avatar
    prashant
    @prashantmital
    Aug 28, 2025
    With Realtime API you can add "voice mode" to your app without introducing server-side complexity. The implementation takes next to no time and many users are already doing a version of this with things like WisprFlow. Seems like a no brainer to me. I am surprised more products
    user avatar
    OpenAI Developers
    OpenAI
    @OpenAIDevs
    Aug 28, 2025
    The Realtime API is officially out of beta and ready for your production voice agents! We’re also introducing gpt-realtime—our most advanced speech-to-speech model yet—plus new voices and API capabilities: 🔌 Remote MCPs 🖼️ Image input 📞 SIP phone calling ♻️ Reusable prompts
    00:00
    721