Peter (@LordMarket22) header
Peter (@LordMarket22) profile picture

Staff AI engineer in fintech. I build AI systems and analyze algorithms, AI harnesses, and businesses: what works, what doesn't, and what it means for investors

Europe
Joined January 2015
Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)Photo by Peter (@LordMarket22)
🚨 GEMINI USERS!! Gemini 2.5 Pro/Flash/Flash-Lite are RETIRING. Your prod needs to move. Like, NOW 😂 We spent the WHOLE weekend moving 9 AI features in our startup off Gemini 2.5, tested on real data: → Feature 1: accuracy (by money) 52% → 81%, 2× faster (ask why and I share) 🤯 → Feature 2: 91.5% → 93%, 2× faster → Feature 3: 3.8 Flash matched 2.5 PRO → Feature 4: Flash Lite, 3× faster → Features 5–6: same answers, swap the model GOTCHAS 👇 → thinkingBudget is deprecated on Gemini 3. Use thinkingLevel → You CAN'T turn thinking off on 3.8 Flash. Even LOW still thinks → Thinking uses up maxOutputTokens, so small caps cut answers off → Smarter model = different behaviour. Our AI judge marked 2× more transactions as "unknown" → Flash Lite is a BIG WIN
10
2
27
5,003
How the fuck did OpenAI go from the worst limits company to the best overnight? 🤔
Day 25/ We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today. We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage. And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
10
1
9
295
🚨 Codex users and friends! I never thought I would see this ERROR message :(. I was EVEN flexing that in my location, the SERVERS are super SAFE and cool. What is the solution, guys? Just wait and retry randomly? Did you get hit by this too? Still at 32% weekly quota. And I need my Astra NOW :D
3
3
163
If you are in tech, let's connect. I am mostly focusing on: 🤖 AI 💹 Investments 🍎 Health 🧮 And learning social algorithms Besides that, I am a staff applied AI engineer at a startup during the day. And I love to create videos about my experiences and interests in my free time I was lucky to gain almost ~140 new and amazing connections last time. Let's see if we can top this up this time.
75
1
55
1,672
Building in public. Showing up every day. Pushing through when you are tired, sick, or have mental or physical health problems or family problems. IS NOT EASY!!! My friend is talking about this, and I became literally EMOTIONAL listening to him and watching this video :( EVERYONE who's struggling or in PAIN, you need to know you are not alone; many of us are going through this! Do not give up, my friend, or anyone else going through this. Also, let's support each other!
Day 30 of building Briwill in public, build in public we were told but no one talks about the mental trauma that comes along with it
4
1
9
283
If you are in tech, let's connect. I am mostly focusing on: 🤖 AI 💹 Investments 🍎 Health 🧮 And learning social algorithms Besides that, I am a staff applied AI engineer at a startup during the day. And I love to create videos about my experiences and interests in my free time Let's start the week by connecting, reconnecting with like-minded people! 🤗
117
2
99
3,645
X-Ray Continues!!! THIRD Episode is HERE. What do X's AI judges decide? 🤖 How many AI judges, and what are they looking for? 🧹 What is a slop grade? 🏷️ How do you get topic labels like News or AI? 🎣 Bait, like begging for likes or reposts? 💬 What happens with your replies? 🪤 Reply to your own reply under a post from a 1,000+ follower account- will it trigger something? Follow for episode 4: Who will your post reach? P.S. Thank you for your patience, everyone!
7
3
1
19
803
Since this post a lot has changed. I have made a lot of friends in a very less time, people highly active in AI Engineering learning and developing harnesses and its engineering, Devops and systems side of things. Machine Learning Audio Engineering Inference Engineering Lot of cool spaces to be in right now. Would want to connect with loads of more people. Haha , enough yap lets larp together :)
Gained 60 amazing connections today and yesterday, I Love the fact connecting with people working in the same niche in tech. Also the thing is friends I have made in just a month living far and wide away but still are a part of journey. Looking to make more genuine connections and helping each other out in our journey to greatness.
8
1
19
513
My friend is hiring and I cannot recommend more if someone is looking for a job, which eventually can make the world a bit better place (faster and more reliable delivery)
Day 26 of building Briwill, Briwill is currently hiring. We’re building a team for our e-commerce company. Know your skill and think you can make Briwill stronger? Dm us with your proposal
2
1
7
260
🚨 codex users!! Tibo gave another RESET for 10 AM PST 🔥 I still had 90% left!!! 😭 Here is YOUR timezone 👇 🇮🇳 India 11:30 PM 🇨🇳 China 2:00 AM 🇮🇩 Indonesia 1:00 AM 🇳🇬 Nigeria 7:00 PM 🇩🇪 Germany 7:00 PM
Replying to @thsottiaux
Lands tomorrow before 10am PST.
16
5
2
47
21,463
Thank you, Anthropic! 🙏 Finally! I have been waiting for this since I first started learning about and using AI. THIS is what I dream of seeing AI used for 👇 → Preventing diseases → Helping people manage chronic conditions → Helping researchers find treatments, and hopefully cures, for diseases we still can’t treat This feels like a step toward that future. I build with AI every day, and health is a big personal focus for me. So seeing AI applied here means a lot. Better tools are not bad, but helping people live longer, healthier lives? Is there a better use case for AI?
We’ve acquired Orchestra Bio, a platform that helps biopharma teams coordinate the complex process of drug development. Together, we’ll build on Rosalind, our AI models and tools for life sciences research, to streamline those operations and help new treatments reach patients sooner. Welcome @ashokaraj_ and team! openai.com/index/openai-to-a…
4
2
10
393
if you are into tech lets connect !!!
28
1
23
630
If you are in tech, let's connect. I am mostly focusing on: 🤖 AI 💹 Investments 🍎 Health 🧮 And learning social algorithms Besides that, I am a staff applied AI engineer at a startup during the day. And I love to create videos about my experiences and interests in my free time The weekend is the BEST time to connect!
25
2
23
828
Hey 👋 Looking for people interested in: - SaaS - Vibecoding - AI Games - Quant Finance - Startups - Solo Founders Drop 🤝 and let's connect. #connect #buildinpublic
52
1
1
28
578
🔬 Deep Dive Inside the USD rates vol desk: Part 3 Swaptions and caps/floors Enjoy the 3rd part of my article series. Informative and hands on!
4
2
1
21
965
Hey friends and strangers, Sorry for going a bit radio silent; crazy work deadlines these past weeks. Still pushing the X stuff, though: → X-Ray under editing → Codex 28-day challenge summary video for @OpenAI Codex → Weekly AI + market videos (incl solana:nasd9ZCaNtC3wZ4G7updmA48VGmcgyNnXsYQ6cRLkjP) None forgotten. Planning to drop them roughly every 2 days. How’s everyone doing? Busy with work or just grinding on X? Hope you’re all well!
5
1
7
346
Another nice addition from Claude. As someone who maintains AI systems serving many customers every day, my take is simple this: Observability is essential in production. You can vibe code your way to an MVP. But once real customers depend on it and money is on the line, the game changes. In my experience, no other platform comes close to Claude when it comes to building stable, robust, maintanable AI systems.
Now in public beta, Claude Platform can send OpenTelemetry traces from the Messages API, Claude Managed Agents, and the Claude SDKs to your observability stack.
2
5
192
One of the first visual demos from my friend @shownotover. What do we think? To be fair Claude Opus had advantage here as Argon run in browser and Opus run in Claude CLI. Still visuals are day and night!
🚨 CLAUDE AND GEMINI USERS!! 👽🍔 The story and the prompt: Aliens stop for burgers… but they LOVE the food on Earth too much 😂 A 15-second Three.js short. Both HIGH. GEMINI ARGON (browser chat) vs CLAUDE OPUS 5.5 (claude cli + toolings) → TIME: ~8–10 min vs 31m54s → TOKENS: Argon ~27.6K vs Opus 5.35M → API COST: Argon ~$0.92 vs Opus ~$6.63 → SPEED: Argon ~46–57 TPS vs Opus ~97.2 TPS WEEKLY: Argon was around ~1% vs Opus close to 0%!!! Argon finished around 3× FASTER, but I still like OPUS’s visual quality WAY more… We need a rematch in Antigravity when we poor Gemini Pro users get access to the CLI version 😂 Which one tells the story better? 👀 How do you like the new Gemini model?
1
1
6
135
🚨 Claude users and Gemini users!!! I’ve only tried Gemini Argon through chat, and I’m NOT convinced so far... Are we really calling this the best model ever??? What’s your experience? Help me out here pls!
29
2
2
27
2,666
I try to answer this, but have no idea. I remember Astra had the lead for some short days 😂 And then suddenly, based on tests: -> all Claude models just wiped the floor with Codex same-tier models. And let's not talk about limits; in a few hours, I use up almost all my limits, really. And do not even mention SPEED (TPS). Wdyt guys?
How the f*ck did Claude suddenly get 100x better than Codex???
1
3
219