b9704

ggml-org/llama.cppb9704Jun 18, 2026by github-actions[bot]same day as b9703

What changed since b9703

low risk

Patch release making the server reject invalid grammars with HTTP 400 instead of silently ignoring them.

New features

  • Server now returns HTTP 400 when a grammar fails to parse, surfacing the error to the caller.

Improvements

  • Grammar parse failures throw explicitly rather than being swallowed, making constraint handling more predictable.
  • Added a regression test covering the invalid-grammar HTTP response.

Bug fixes

  • Fixed issue #24144 where invalid grammar constraints were silently dropped and requests proceeded unconstrained.

Upgrade notes

Clients that previously sent malformed grammars and received a normal completion will now get HTTP 400, so add error handling for that status in any grammar-constrained request path. No configuration or data migration is required; the -hf preset changes from b9703 are unchanged here.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

A build release fixing server error handling for invalid grammar.

Key Highlights

  • Return HTTP 400 on invalid grammar instead of silently dropping constraints
  • Added regression test for invalid-grammar response

Release Notes (excerpt)

<details open> server : return HTTP 400 on invalid grammar (#24144) (#24154) Throw on grammar parse failure so the server returns HTTP 400 instead of silently dropping the constraint. Add a regression test for the invalid-grammar response. Fixes #24144 </details> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories