deps(backend)(deps): bump llama-cpp-4 from 0.2.61 to 0.3.0 in /backend

[dependabot]

Bumps llama-cpp-4 from 0.2.61 to 0.3.0.

Release notes

Sourced from llama-cpp-4's releases.

v0.3.0

Summary

Minor release bumping the vendored llama.cpp submodule, expanding the MTP Rust API, and fixing Linux dynamic prebuilt collection in CI.

Changed

• llama.cpp: bumped vendored submodule to b28a2f372 (includes MTP clean-up #23269).
• MTP draft API: MtpSession::new_with_config and MtpSessionConfig expose n_min and p_min; upstream default p_min is now 0.0.
• CI: Linux dynamic prebuilt collection now includes versioned .so files and symlinks.

Added

• MtpSession::need_embd_pre_norm, MtpSession::print_stats, MtpSession::config, MtpSession::n_min, MtpSession::p_min.
• examples/mtp: --p-min CLI flag; session stats printed after generation.
• Expanded MTP documentation in README, MTP.md, and rustdoc.

Crates

• llama-cpp-4 0.3.0
• llama-cpp-sys-4 0.3.0

Full changelog: https://github.com/eugenehp/llama-cpp-rs/blob/v0.3.0/CHANGELOG.md#030---2026-05-19

Changelog

Sourced from llama-cpp-4's changelog.

[0.3.0] - 2026-05-19

Changed

• llama.cpp: bumped vendored submodule to b28a2f372 (includes MTP clean-up #23269).
• MTP draft API: MtpSession::new_with_config and MtpSessionConfig expose n_min and p_min; upstream default p_min is now 0.0.
• CI: Linux dynamic prebuilt collection now includes versioned .so files and symlinks.

Added

• MtpSession::need_embd_pre_norm, MtpSession::print_stats, MtpSession::config, MtpSession::n_min, MtpSession::p_min.
• examples/mtp: --p-min CLI flag; session stats printed after generation.

[0.2.56] - 2026-05-16

Changed

• llama.cpp: bumped vendored submodule to 64b38b561 (master, 2026-05-16), which now includes upstream MTP support (PR #22673, COMMON_SPECULATIVE_TYPE_DRAFT_MTP / LLAMA_CONTEXT_TYPE_MTP).
• Patch removed: llama-cpp-sys-4/patches/0002-mtp.patch is gone — its functionality is now upstream. The mtp Cargo feature has been removed from both llama-cpp-sys-4 and llama-cpp-4.

Added

• New LlamaContextType { Default, Mtp } enum and LlamaContextParams::with_ctx_type / ctx_type wrapping upstream's llama_context_type (use Mtp to load a draft head as the MTP context for upstream's --spec-type draft-mtp speculative decoder).
• LlamaContextParams::with_n_rs_seq / n_rs_seq and LlamaContext::n_rs_seq are now always available (no feature gate).
• New llama_cpp_4::mtp::MtpSession — Rust-callable MTP speculative-decoding draft loop. Wraps a small C++ shim (llama-cpp-sys-4/mtp_shim/mtp_shim.cpp) that re-exports upstream's common_speculative_* MTP path with stable C linkage. Smoke-tested end-to-end on Qwen3.6-27B-IQ2_M with 94% draft acceptance.
• New examples/mtp/ — without --predict configures contexts (smoke test); with --predict N drives the full draft loop via MtpSession.

Removed (breaking)

• mtp Cargo feature on both crates.
• LlamaContext::set_mtp — upstream removed the llama_set_mtp C API; MTP is

... (truncated)

Commits

• d7136d7 0.3.0
• See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

[pacphi]

✅ Snyk checks have passed. No issues have been found so far.

Status	Scan Engine	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.