Models & APIs

Qwen 3 Coder Plus available through standard SkillBoss access.

Qwen 3.5 Flash available through standard SkillBoss access.

Qwen 3.5 Plus — Alibaba's native vision-language Plus model featuring a hybrid architecture that fuses linear attention with sparse Mixture-of-Experts, delivering superior inference efficiency. Supports multimodal inputs including text, images, and video with advanced visual understanding and deep reasoning. Demonstrates a significant performance leap over Qwen 3 in both pure-text and multimodal tasks, matching current frontier models. Built-in support for function calling, structured output, web search, prefix completion, and prompt caching.

Qwen Coder Plus available through standard SkillBoss access.

Qwen Flash: Advanced AI chat model for natural language processing, text generation, and conversational AI tasks. Integrate with Claude Code, Cursor, and other AI agents for automated workflows.

Qwen Max: Advanced AI chat model for natural language processing, text generation, and conversational AI tasks. Integrate with Claude Code, Cursor, and other AI agents for automated workflows.

Qwen Plus: Advanced AI chat model for natural language processing, text generation, and conversational AI tasks. Integrate with Claude Code, Cursor, and other AI agents for automated workflows.

Qwen VL Max available through standard SkillBoss access.

Qwen VL Plus available through standard SkillBoss access.

Qwen3.7 Max, Alibaba's current flagship Qwen model, with thinking and non-thinking modes and up to 1M-token context.

Upscale an image through the bounded Replicate Real-ESRGAN adapter.

Use retrieve-knowledge to extract and process document information for AI agents. Ideal for Claude Code automation, data extraction workflows, and intelligent document analysis tasks.

Seedance 2.0 — high-quality text/image/video-to-video generation at 480p / 720p / 1080p, up to 15 seconds. Asynchronous: pass {"async": true} for a 1-second submit + poll status.

Seedance 2.0 Fast — cost-effective variant, ~20% cheaper at equivalent parameters. 480p / 720p (no 1080p), up to 15 seconds.

Real-time Bing organic SERP for keyword + location.

Real-time Google Maps local pack results: business names, ratings, reviews, phone, hours, place_id.

Real-time Google News results: article headlines, publishers, timestamps, snippets.

Real-time Google organic SERP for any keyword + location. Returns ranked URLs, titles, snippets, sitelinks, knowledge graph. Up to 100 results per request.

Real-time YouTube search results: video titles, channels, views, durations, thumbnails.

Shotstack Edit — programmatic video editing via JSON timeline. Submit a `timeline` (tracks, clips, transitions) and `output` spec (format, resolution, fps), receive the rendered video URL. Supports video, image, audio, text, and HTML clips. Billed per second of rendered video duration.

Shotstack Ingest — fetch a media file from a public URL and make it available as a Shotstack source for use in render timelines. Returns a source id that can be referenced as a clip asset. Billed per request.

Inspect a public media URL through the Shotstack probe contract.

Shotstack Probe — inspect a media file (video, audio, or image) by URL and return its properties: codec, resolution, bitrate, duration, streams, and container format. Synchronous. Billed per request.

Read the status of a Shotstack render by render ID.