This PR aligns multimodal_gen image input handling with sglang.srt semantics and removes unnecessary remote-image persistence in the OpenAI entrypoints.

What Changed

• Defaulted remote image inputs (URL / data URI) to in-memory loading in OpenAI image/video paths.
• Reused the shared srt-style image loading flow (get_image_bytes-based) in multimodal_gen vision loading.
• Added lightweight retry logic for URL image download to handle transient network/HTTP failures (e.g. timeout, 429, 5xx).

Why

• Avoid save-then-load overhead for remote image inputs.
• Reduce disk I/O pressure and temp-file dependency in hot paths.
• Improve robustness for unstable remote image fetches.

Motivation

Modifications

Accuracy Tests

Speed Tests and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review and Merge Process

1. Ping Merge Oncalls to start the process. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • Common commands include /tag-and-rerun-ci, /tag-run-ci-label, /rerun-failed-ci
4. After green CI and required approvals, ask Merge Oncalls or people with Write permission to merge the PR.