Archive, page 3
- Jev's speed numbers hold up, Theo says not for compaction
Guillermo Rauch's team measured Jev up to 18x faster at p95 than GPT Luna on a safety classifier, and more accurate. The same week, Theo called using Jev to compact agent history a terrible strategy, which is the argument worth reading.
- An AI-pilled company capped its SOTA model budget
Gergely Orosz talked to a company that had unlimited AI budgets a year ago and a technical CEO bullish on AI. It now runs a daily cap and uses the priciest models for planning only.
- Theo vs the Jev compaction trick
A one second compaction hack reportedly took a Claude session from nearly 1M tokens to 86K, and Theo called it a terrible strategy within hours. Armin Ronacher told him not to dismiss it, then someone benchmarked it and sided with Theo.
- Jev has everyone rewriting their classifiers
Malte Ubl says Jev beat Gemini 2.5 Flash Lite on an existing classifier eval, winning on quality and running 6x faster, and Mario Zechner says it replaces seven years of his own classifier work. The model everyone tested today is not a chat model.
- Mitchell Hashimoto's whiteboard test for AI code
Gary Bernhardt says his reviews routinely cut an agent's diff to 25% of its original size, in the same week Mitchell Hashimoto and Boris Cherny set out what you actually have to understand before shipping. The argument moved from whether agents can write the code to who is accountable for it.