hermes - 💡(How to fix) Fix Gemini native adapter should accept OpenAI-format video_url parts for video analysis

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Fix Action

Fix / Workaround

Local workaround tested

A local patch changed _extract_multimodal_parts() to accept both image_url and video_url part types. After this, video_analyze using gemini-3.1-flash-lite successfully described an MP4 test video.

RAW_BUFFERClick to expand / collapse

Bug description

Hermes video analysis routes may pass multimodal content using OpenAI-style video_url parts. The Gemini native adapter's multimodal extraction path handled image_url, but not video_url, so video input could be dropped or fail before reaching Gemini's native inlineData handling.

Steps to reproduce

  1. Configure Gemini as the backend for video analysis.
  2. Call video_analyze on a local/URL MP4.
  3. The tool constructs an OpenAI-style video_url content block.
  4. The Gemini native adapter only extracts image_url parts.

Actual behavior

The video payload is not mapped into Gemini native content correctly, leading to failed or ineffective video analysis.

Expected behavior

The Gemini native adapter should treat video_url as a media URL part, download/encode it like other supported media inputs, and emit Gemini native inlineData with the appropriate MIME type.

Local workaround tested

A local patch changed _extract_multimodal_parts() to accept both image_url and video_url part types. After this, video_analyze using gemini-3.1-flash-lite successfully described an MP4 test video.

Environment

  • Hermes checkout commit: 9f182bd7b
  • OS: Ubuntu 24.04 / Linux
  • Provider/model: Gemini / gemini-3.1-flash-lite

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

FAQ

Expected behavior

The Gemini native adapter should treat video_url as a media URL part, download/encode it like other supported media inputs, and emit Gemini native inlineData with the appropriate MIME type.

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING