b9977

ggml-org/llama.cppb9977Jul 12, 2026by github-actions[bot]same day as b9976

What changed since b9976

low risk

Server fix preserves image blocks in Anthropic tool_result during OpenAI conversion, restoring multimodal tool outputs.

New features

  • Multimodal tool results now survive Anthropic-to-OpenAI conversion, so tools returning images reach the model.
  • Image blocks inside tool_result are emitted as OpenAI multimodal content parts (text plus image_url array).

Improvements

  • Plain-text tool results still serialize as simple strings, keeping backward compatibility with existing clients.
  • Added a regression test covering image blocks in Anthropic tool_result conversion.

Bug fixes

  • Fixed server_chat_convert_anthropic_to_oai() silently discarding image blocks nested in Anthropic tool_result content.

Upgrade notes

This is a drop-in patch with no config or API changes required. Users of the Anthropic-compatible endpoint whose tools return images should upgrade to get correct multimodal behavior; text-only tool flows are unaffected.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

Fixes a bug where image blocks in tool results were silently dropped during the Anthropic-to-OpenAI conversion process, breaking multimodal tool outputs.

Key Highlights

  • Multimodal tool output support
  • Anthropic-to-OpenAI conversion fix
  • Image block preservation in tool results

Release Notes (excerpt)

<details open> server : fix image blocks in tool_result being dropped during Anthropic OpenAI conversion (#22536) * server : fix image blocks in tool_result being dropped during Anthropic→OpenAI conversion server_chat_convert_anthropic_to_oai() silently discarded image blocks inside Anthropic tool_result content. This broke multimodal tool outputs (e.g. a tool that returns an image) because the model never received the image. When tool_result contains image blocks, convert them to OpenAI…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories