Opus 5.5 finished Theo's Rust port after Astra stalled
Theo had been stuck at 85% of tests passing on a four month TypeScript to Rust port. He says Opus 5.5 finished it in 10 hours, then he burned through roughly 2.5 Claude accounts letting it optimize.
The big disagreement
Theo spent $1,000 benchmarking the Jev router and calls it pointless
OpenRouter launched typesafe/jev-router, which uses Jev to pick a model and reasoning effort per request, and Theo benchmarked it for $1,000: on DeepSWE it scored roughly the same as GPT-6 Astra on low, cost slightly more, and took almost 5x longer.
In order to know how much thinking a task requires, you have to think. Jev does not "think".
Receipts
Opus 5.5 finished Theo's four month Rust port in 10 hours
Theo says Opus 5.5 got his ts-rust port working in 10 hours after four months of on and off work, then spent another 24 hours on performance. He had reached about 35% of tests passing with GPT-5.6 Sol and about 85% with GPT-6 Astra, both of which stalled and looped for days.
I can not believe how quickly Opus unblocked the work Astra was stuck on.
DHH had Opus 5.5 translate a Rust screensaver to x86-64 assembler
DHH says he ported the Omarchy screensaver engine (ttfx) from Rust to x86-64 assembler in a one-shot translation by Opus 5.5, up to 17x faster. Overnight it went further: 22x faster at peak, 6x on the mean, and up to 450x faster than the original Python, with SSE2, AVX2 and AVX-512 paths and a fallback to the Rust version.