Topic Page

Benchmarks

Posts connected to this thread.

OpenAI Stops Using SWE-bench Verified For Frontier Coding Claims: What Codex Teams Should Measure Now

February 23, 2026 · 7 min read

The most important Codex-related development on February 23, 2026 is not a new model launch. It is an evaluation policy change. OpenAI published a research post stating …

GPT-5.3-Codex: What The Benchmark Gains Actually Mean

February 20, 2026 · 4 min read

OpenAI introduced GPT-5.3-Codex on February 5, 2026 with two claims developers immediately care about: materially better benchmark scores and faster median response time. …