zig

• 2026-10-16
  The engine got credit for the quant

  open-jet beat llama.cpp 7 to 3 on a Zig agent task. On open-jet's own GGUF, llama.cpp went 7 of 9 too. Only the speed gap belonged to the engine.

  The 3090 Agent Bench · 7/27 11 min read
• 2026-10-15
  Zig needed context, not time

  Four 27B local models kept hitting the one-hour wall on a Zig task. Four hours at 64k got 7 of 12 through the core tests. At 128k, 35 of 36 did, cleanly.

  The 3090 Agent Bench · 4/27 15 min read
• 2026-10-13
  The deadlock test can't see a half-held lock

  Every model wrote a thread test. Most of them could not fail: with locks taken in id order, opposite-direction transfers serialize on one mutex.

  The 3090 Agent Bench · 3/27 13 min read
• 2026-10-12
  Zig's defer ends at the closing brace

  A local model locked two accounts in the right order, scored 97/100, and lost money. One Zig defer sat inside an if block and unlocked at the brace.

  The 3090 Agent Bench · 2/27 9 min read
• 2026-10-12
  I blamed OpenCode; a 2-bit model passed Zig in it

  Four 27B models failed a Zig agent task in OpenCode, and the client looked guilty. Then a 2-bit MoE in the same setup passed 6 of 6.

  The 3090 Agent Bench · 1/27 13 min read