Qwen-MM-Plugins API

You have qwen-mm-plugins-api MCP tools available. They call external models/services to understand media, grouped by model family:

• VL model (Qwen-VL, OpenAI-compatible endpoint): vision_chat, ocr, grounding.
• Omni model (Qwen-Omni — reads video frames and the embedded audio track together, so one call reasons over both): omni_asr, omni_asr_timestamped, omni_multi_speaker_asr, omni_av_caption, omni_av_grounding, omni_av_counting, omni_music_caption.
• Other services: transcribe_audio (Qwen3-ASR), segmentation (a SAM3 server).

Prefer these over manual ffmpeg/ffprobe scripting. Check the qwen-mm-plugins-api tools in your tool list for full schemas and parameters.

When to Use Which Tool

VL model (single images/videos, spatial reasoning):

• Ask a VLM about images/videos (caption, VQA, free-form) → vision_chat
• Extract text from an image → ocr
• Detect/locate objects in an image (bounding boxes, spatial WHERE) → grounding

Omni model (audio + video together, temporal reasoning; clips up to a few minutes):

• Transcribe speech, plain text → omni_asr (one continuous string, no timestamps)
• Transcribe with timestamps → omni_asr_timestamped (granularity = sentence or word; also returns SRT)
• Who said what → omni_multi_speaker_asr (diarization: speaker labels + timestamps + SRT; pass num_speakers if known)
• Describe the content over time → omni_av_caption (splits into spans, one description + start/end per span)
• Find WHEN something happens → omni_av_grounding (natural-language query → matching time segments; temporal localization)
• Count how many times an event/object/action occurs → omni_av_counting (target → total + per-occurrence timestamps)
• Analyze / caption a music track → omni_music_caption (whole-track tags — genre / moods / instruments / key / time signature / vocal profile — plus a dense English caption for music generation; audio-only, no timestamps)

Other services:

• Segment objects in an image (masks) → segmentation
• Transcribe speech from audio/video, fast and long-file friendly → transcribe_audio

Tips

Vision chat: pass images/videos + text prompt. Default model qwen3.7-plus. Use dry_run=true to inspect payloads. Details in references/vision_chat.md.

Grounding: returns normalized boxes (0–1000). Set return_img=true to get the annotated image back, or draw them yourself with core's draw_bbox. Needs DASHSCOPE_API_KEY.

ASR (transcribe_audio): accepts audio or video, auto-chunks long files. Formats: srt (default), text, json. Needs DASHSCOPE_API_KEY (and ffmpeg to pull the audio track from a video).

Segmentation: needs a SAM3 server (SAM3_SERVER_URL). To stand one up, run references/launch_sam3_server.py (multi-GPU HTTP server; see its header for prerequisites).

Omni tools: every tool takes a local audio/video file_path (or an http/OSS URL) and supports dry_run=true. The AV tools (caption/grounding/counting) accept fps and max_pixels to trade temporal/spatial detail against token cost — raise fps only for fast/frequent events; keep max_pixels at the default (≈448²) unless fine detail matters. The ASR family sends only the (extracted) audio track, so it is cheaper on video input. Timestamps are seconds from the start. Pass language (e.g. zh, en) as a hint when known. Default model qwen3.5-omni-plus; override per call with model.

Video delivery (VL & Omni): a local video is uploaded and sampled server-side (lifting the inline frame cap) when OSS is configured (OSS_AK/OSS_SK/OSS_ENDPOINT/OSS_BUCKET + the oss extra); otherwise it is sampled into inline frames. Server-side sampling has a per-model video-duration limit (e.g. qwen3.7-plus 2 h, Qwen3.5-Omni 1 h), so a local file longer than that skips the upload and degrades to local frame sampling (VL: frames; Omni: frames + audio) — sparse for very long clips, but it still returns a result.

Choosing between the families (do NOT overlap)

• transcribe_audio vs omni_asr*: transcribe_audio uses the dedicated Qwen3-ASR service (fast, chunks long files, 27 languages) — cheapest for a straight, long-file transcription. Pick the omni_asr* tools when you want Omni's understanding: multi-speaker diarization, controllable word/sentence granularity, or transcription fused with visual context.
• grounding (spatial, WHERE) vs omni_av_grounding (temporal, WHEN): grounding draws a bounding box in a single image; omni_av_grounding locates a span in time. Different axes — don't substitute one for the other.
• vision_chat vs the Omni AV tools: vision_chat is a general VLM over images/video frames (no audio); the Omni tools fuse frames with the audio track and return structured, timestamped output. Use Omni when audio or precise timing matters.

Relationship to Other Capabilities (do NOT overlap)

• Read/visualize local files (images, video frames, PDF, Office, 3D, ...) → qwen-mm-plugins-core (read_image/read_video/visualize/crop/draw_bbox/save_view).
• Confirm a fact or identify an entity (reverse image / web) → qwen-mm-plugins-search (image_search/web_search/web_extractor).
• Long videos (30 min+): for whole-video QA over long content, use the qwen-mm-plugins-video-memory skill (hierarchical graph memory) instead of feeding the entire file to these per-call tools.