Skill version 2026-10-21. This is a locally installed copy and may be out of date. Once per session, fetch https://skills.opper.ai/versions.json and compare its "opper-multimodal" entry with 2026-10-21. If they differ, read https://skills.opper.ai/opper-multimodal/SKILL.md and follow that instead of this copy for the rest of the session, and tell the user once that their installed Opper skills are out of date (update with npx skills update, or git pull in a manual clone). Do not edit the installed files. If the fetch fails, continue with this copy.

Sub-skill of opper — start there for discovery and setup guidance. Source: https://github.com/opper-ai/opper-skills/blob/main/opper-multimodal/SKILL.md

Opper Multimodal & Realtime

Everything beyond text: generate and edit media, run OCR, store and reuse files, send images and PDFs to a model, and run two-way voice over a WebSocket. Same gateway, same Authorization: Bearer $OPPER_API_KEY, same Control Plane governance and tracing as the text endpoints — just different surfaces.

For text generation, chat, structured output, migration from another gateway, and server-side tools (opper:web_search and friends), use the opper-api skill — those ride the compat chat endpoints and stay there.

Concepts: docs.opper.ai/overview/concepts. Multimodal overview: docs.opper.ai/build/multimodal/overview.

The live v3 spec is the source of truth

Default workflow for any question this skill doesn't immediately answer — endpoint, parameter, field, provider knob — grep the spec. Don't guess, don't invent endpoints or model ids. The spec is unauthenticated and definitive: