martianzhang

Use "aigc-cli midjourney" (alias "mj") to generate and edit images via Midjourney on APIMart. Supports imagine, blend, describe, edits, upscale, variation, reroll, zoom, pan, inpaint, modal, video, remix. All async task-based. Automatically polls task and downloads results.

Use "aigc-cli image" to generate images via OpenAI-compatible APIs (APIMart, OpenAI, OpenRouter). Supports text-to-image, image-to-image, inpainting, Grok Imagine Edit (--edit), local file upload, dry-run, proxy, sync/async mode detect. Automatically polls task and downloads images.

Use "aigc-cli video" to generate videos via the APIMart API (doubao-seedance-2.0, VEO3). Supports text-to-video, image-to-video, first/last frame, reference video/audio, audio-enabled video, VEO3 Remix (--remix), seed, web_search tool, and return-last-frame for continuation. Automatically polls task and downloads videos.

Use "aigc-cli image" with OpenRouter to generate images via the Responses API. Supports text-to-image with aspect ratio control. Automatically detects OpenRouter from --api-base / OPENAI_BASE_URL and uses POST /v1/responses with image output modalities.

Use "aigc-cli video" with OpenRouter to generate videos via the dedicated video API. Supports text-to-video, image-to-video (first frame), aspect ratio, resolution, and duration control. Uses async submit → poll → download pattern.

Convert table data between different formats (MySQL, CSV, JSON, Markdown, HTML, SQL, Excel, XML, LaTeX, MediaWiki, TWiki, templates)

Large Language Model Capability Testing Framework — Systematically evaluates LLMs on mathematical reasoning, commonsense judgment, causal logic, character recognition, kinship relations, and local knowledge. Suitable for testing GPT, Claude, Gemini, Qwen, DeepSeek and other mainstream LLMs.