Build software
Reason across specifications and code, then call tools to move from plan to implementation.
Reason over code, documents, and images with GPT-6 Astra—then use imageat MCP to generate visuals, video, and production-ready creative assets.
Reason across specifications and code, then call tools to move from plan to implementation.
Turn a brief into image, video, editing, and campaign tasks through imageat MCP.
Include image URLs alongside text for visual review, extraction, and multimodal decisions.
Enable per-request web search when the answer depends on current information.
Use OpenAI-format tools, tool choice, parallel calls, and structured tool results.
Choose low, medium, high, xhigh, or max effort and tune answer verbosity separately.
Made with Astra + imageat MCP
Start with an ambitious brief. Astra plans the work and calls imageat tools to turn each stage into real creative output.
Astra plans the shot, then calls imageat MCP to bring the miniature city to life — ferries crossing the Bosphorus, rippling water, and window lights coming up across the hills.
Use imageat MCP to create a short cinematic video using the appropriate Wiro-powered video model. Scene: a museum-grade miniature of Istanbul built on a dark machined display table, seen from a low three-quarter angle at night. The Bosphorus and the Golden Horn cut through the model in deep blue; the historic peninsula, Galata Tower, Hagia Sophia, the Blue Mosque and dense terracotta rooftops fill the land, and thin lime-green light guides trace the coastline, the roads and the edge of the table. Motion: a slow five-second forward push that drifts a few degrees to the right and rises gently, horizon kept level, no shake or whip. Ferries and small boats glide along the strait leaving soft luminous wakes, the water ripples with fine parallax, traffic crawls over the suspension bridge, and window lights switch on in staggered waves from the waterfront up the hills. Lighting: night key from the miniature's own emissive windows, cool ambient fill, a subtle green rim from the table lighting, soft volumetric haze above the water, and a tilt-shift depth of field that keeps the centre of the model sharp. Preserve landmark placement, architectural proportions, table geometry and overall composition throughout the shot — nothing may morph, drift off the base, or change scale. Inspect the generated result and retry or refine if there are visible continuity, deformation, lighting, or motion issues. Deliver the final video.
One brief becomes a mechanically coherent exploded view. Every lens element, board and screw travels on its own clean axis and holds its scale.
Use imageat MCP to create a short cinematic video using the appropriate Wiro-powered video model. Scene: a professional cinema camera rig suspended against a black studio backdrop and separated into a technical exploded view — body shell, sensor block, circuit boards, cooling fan, side monitor, top handle, baseplate, 15mm rods, cine prime barrel, lens mount ring and the full stack of glass elements floating in line, with individual screws hanging in the air around them. Thin lime-green light ribbons sweep through the frame behind the parts. Motion: a controlled five-second move in which the components ease further apart along clean straight axes at slightly different speeds and then settle and hold, while the camera performs a slow lateral orbit with a subtle dolly-in and a focus rack from the front lens element back to the sensor. Lighting: hard studio key with soft rim light on every part, crisp specular streaks across machined aluminium and anodised black metal, magenta and green coating flare on the lens elements, and clean reflections with no blown highlights. Preserve mechanical coherence, relative part scale and assembly logic throughout the shot — every component keeps its identity and stays on its own axis, and nothing merges, duplicates, dissolves or rotates out of alignment. Inspect the generated result and retry or refine if there are visible continuity, deformation, lighting, or motion issues. Deliver the final video.
Three linked shots of the same model, cut into a premium commercial. Identity, wardrobe and colour grade hold steady from the close-up to the rooftop stage.
Use imageat MCP to create a short cinematic video using the appropriate Wiro-powered video model. Scene: a high-fashion commercial built from three linked shots of the same young woman with a blunt black fringe. Shot one is an extreme close-up on her eyes as she opens them and city reflections catch in her irises. Shot two is a top-down frame of her standing perfectly still on a wide city crossing in a cream structured jacket over black, while commuters stream past her in motion blur. Shot three is a night rooftop stage where she and three dancers hold a formation under crossed spotlight beams with a neon skyline behind them. Motion: one deliberate camera move per shot — a slow push in on the close-up, a steady rise over the crossing, a slow arc around the rooftop stage — and performers who move with controlled choreographed weight rather than drifting or sliding. Lighting: premium commercial grade, with soft daylight on the street shots, a hard key with crossed beams and volumetric haze on the rooftop, and cool teal and magenta accents held consistent across all three shots. Preserve facial identity, hair length and fringe shape, makeup, and the exact garment cut and colour across every shot. Keep hands, fingers and limb proportions anatomically correct, and avoid frame-to-frame flicker, warping faces, extra or merging limbs, sliding feet, and any wardrobe change mid-shot. Inspect the generated result and retry or refine if there are visible continuity, deformation, lighting, or motion issues. Deliver the final video.
Live playground
Run the model, inspect its schema, then move the same request into production with one imageat API key.
Sign in to use the live model.
Your response will appear here
No plugin required
Add the remote MCP URL to ChatGPT, Claude, Cursor, VS Code, or another compatible client. Sign in to imageat during OAuth, then ask Astra to use imageat tools for image, video, and editing work.
{
"servers": {
"imageat": {
"type": "http",
"url": "https://mcp.imageat.com/mcp"
}
}
}Developer API
Use the OpenAI-compatible Chat Completions endpoint with your imageat API key. Astra accepts role-based messages, image URLs, reasoning and verbosity controls, web search, and function tools.
curl -X POST "https://api.imageat.com/v1/chat/completions" \
-H "Authorization: Bearer $IMAGEAT_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-6-astra",
"messages": [{"role":"user","content":"Plan and build a polished product launch."}],
"reasoning": "high",
"verbosity": "medium",
"web_search": true,
"tools": [{
"type": "function",
"function": {
"name": "create_campaign_asset",
"description": "Create an imageat campaign asset",
"parameters": {"type":"object","properties":{"prompt":{"type":"string"}},"required":["prompt"]}
}
}]
}'Prototype in the browser, connect MCP to your AI client, or ship through the API.