Tech News
← Home  ·  All topics

Fuzzing

3 GoKawiil briefs on this topic · RSS for fuzzing

GPT-6 Astra outperforms GPT-5.6 Sol on Zstd and IMAP coding evals

An evaluation comparing GPT-6 Astra to GPT-5.6 Sol on software test/verification tasks found Astra achieved over 99% full test correctness on a Zstd implementation task at medium effort for about $1, versus Sol's 63-77% correctness at $3.50-$5.50. The Zstd eval has become saturated for Astra, nearly maxing out regardless of testing technique used, so evaluators also tried an IMAP RFC implementation task that had been too difficult for Sol to attempt meaningfully.

Researcher outlines manual and automated methods for testing kernel race condition bugs

A security researcher describes the practical difficulty of confirming and reproducing race condition bugs in software, particularly the Linux kernel. Current techniques include manually inserting delay calls into kernel code to force specific thread interleavings, or using DTrace probes on macOS and Windows, both of which are slow and require trial and error.

Developer builds structure-aware fuzzer to hunt bugs in the Gleam compiler

A developer frustrated by inconsistent Erlang and JavaScript outputs from the Gleam compiler first tried using an LLM to generate edge-case test programs, spending about $20 in tokens to surface just one confirmed bug (issue #5613), which was promptly fixed. Dissatisfied with the cost and randomness of that approach, they pivoted to building a structure-aware fuzzer that generates syntactically valid Gleam programs automatically to systematically probe the compiler for hidden defects.