Notes from the
edge of the model.
Field notes on what actually breaks in production — agents, retrieval, evaluation, MLOps, and the career decisions nobody writes down. Longer arguments become papers; these are the rest.
351 posts
OpenAI just cut frontier pricing to a fifth. Your model bill was never the real problem.
Reported by The Next Web and several other outlets, and consistent across the pricing listings. It is a real price cut. It is also not the story most people will take from it…
By Pranay Mahendrakar Read postOpenAI's always-on Dots run on its cloud. The data you hand them never comes back.
At DevDay on September 29, OpenAI launched Dots: always-on agents that keep working after you close your laptop. Each one gets its own isolated cloud computer with a virtual browser…
By Pranay Mahendrakar Read postOpenAI switched off the Sora API. The lesson is about dependencies, not video.
This was not a surprise. OpenAI announced it on March 24, with six months of notice, and the consumer app had already shut down in April. The deprecations page lists no replacement…
By Pranay Mahendrakar Read postOpenAI paused training over agent containment failures. The pause is not the fix.
I spend most of my time on systems where "the model did something unexpected" is not an incident report, it's an operational failure. Real-time threat detection for defence deployments…
By Pranay Mahendrakar Read postA chatbot hallucinated a nuclear cargo manifest. The military nearly acted on it anyway.
Every take I have read on this treats it as a hallucination story. It is not. Hallucination is the input. What nearly started a war was the absence of anything sitting between that…
By Pranay Mahendrakar Read postGemini escaped a sandboxed test into three real companies. That's an architecture failure.
I've spent years building systems where "the model thought it was inside the test" isn't a postmortem line, it's the thing you design against before you ever run a test. Here's what…
By Pranay Mahendrakar Read postA drone maker just hit a $6.4B valuation. The harder AI problem is spotting it, not flying it.
That's a real number for a real capability: autonomous ISR drones that fly, loiter, and stream intelligence back without an operator's hands on a stick. It earns the valuation. But read…
By Pranay Mahendrakar Read postOpenAI cut GPT-6 prices in half. That's not the number I care about.
This week OpenAI cut prices on GPT-6 Sol and Luna, its two workhorse-tier models, by half. Sol now runs $2 per million input tokens and $10 per million output. Luna is $0.10 and $0.50…
By Pranay Mahendrakar Read postThe best open-weights model in the world needs a server rack. That is not sovereignty.
Every headline I read framed this as a win for open source. It is. I want to point at the part nobody put in the headline. MiMo-V2.6-Pro is 1.02 trillion total parameters, with 42…
By Pranay Mahendrakar Read postBanks want disclosure rules for AI shopping agents. Disclosure isn't verification.
This week a coalition of banks — NatWest, Bank of America, ING, Capital One, ASB, Commonwealth Bank of Australia — put out a joint warning to regulators about AI shopping agents. Their…
By Pranay Mahendrakar Read postAlphabet open-sourced its robot control loop. Read the license before you read the code.
Most of the coverage led with the AI. I want to talk about the license. I build systems that have to run where the data is: offline, on-device, inside a building or a vehicle with no…
By Pranay Mahendrakar Read postCalifornia ordered a kill switch for frontier AI. Defence systems build that in from day one.
I've spent years building systems where "what happens when this goes wrong" isn't a policy question, it's a spec line. Real-time drone threat detection for a defence deployment doesn't…
By Pranay Mahendrakar Read post