AI Models API

Text and image generation over a single HTTP endpoint. No SDK, no keys to pass — just hit the URLs below.

\n\n

\n This API gives you two things: a chat endpoint that returns text from a language model,\n and an image endpoint that turns a text prompt into an image and hands you back a hosted URL.\n Everything speaks JSON. You can call it from a browser, curl, a phone app, or any backend.\n

\n\n

\n There's no auth on these routes — the server holds the model token internally, so you only\n deal with prompts and responses. If you're building a public app, put your own rate limiting in front of it.\n

\n\n

Base URL

All routes are relative to the deployment origin:

loading…

\n\n

Quick start

The fastest way to see it work — open this in your browser:

loading…

\n\n \n

Chat

Send a prompt, get text back. There are two ways to call it.

\n\n

\n GET\n /v1/chat?prompt=...\n

Good for quick tests and simple one-liners. Put your question in the prompt query parameter.

curl \"$BASE/v1/chat?prompt=What%20is%20the%20capital%20of%20France%3F\"

\n\n

\n POST\n /v1/chat\n

Use this when you want a system prompt, a full conversation, or generation settings. Send JSON.

\n\n

Body fields

\n \n \n \n \n \n \n \n \n

Field	Type	Notes
`prompt`	string	Shortcut for a single user message. Use this or `messages`.
`messages`	array	Full chat history. Each item is `{ \"role\": \"system\" \| \"user\" \| \"assistant\", \"content\": \"...\" }`.
`model`	string	optional · defaults to `openai/gpt-oss-120b:groq`
`temperature`	number	optional · 0–2, higher is more random
`max_tokens`	number	optional · cap on response length
`top_p`	number	optional · nucleus sampling

\n\n

Example — single prompt

curl -X POST \"$BASE/v1/chat\" \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"prompt\": \"Give me three names for a coffee shop.\"\n  }'

\n\n

Example — full conversation

curl -X POST \"$BASE/v1/chat\" \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"model\": \"openai/gpt-oss-120b:groq\",\n    \"temperature\": 0.7,\n    \"messages\": [\n      { \"role\": \"system\", \"content\": \"You answer in one short sentence.\" },\n      { \"role\": \"user\", \"content\": \"Why is the sky blue?\" }\n    ]\n  }'

\n\n

Response

{\n  \"model\": \"openai/gpt-oss-120b\",\n  \"reply\": \"Sunlight scatters off air molecules, and blue light scatters most.\",\n  \"usage\": {\n    \"prompt_tokens\": 24,\n    \"completion_tokens\": 14,\n    \"total_tokens\": 38\n  }\n}

\n\n \n

Image

\n Describe an image, get back a hosted URL you can drop into an <img> tag.\n The server generates the image, uploads it for you, and returns the link.\n

\n\n

\n GET\n /v1/image?prompt=...\n

\n POST\n /v1/image\n

\n\n

Body / query fields

\n \n \n \n \n \n \n

Field	Type	Notes
`prompt`	string	What you want to see. Required.
`model`	string	optional · defaults to `Tongyi-MAI/Z-Image-Turbo`
`negative_prompt`	string	optional · things to avoid in the image
`raw`	boolean	optional · if true, returns the raw image bytes instead of a JSON URL

\n\n

Example

curl -X POST \"$BASE/v1/image\" \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"prompt\": \"An astronaut riding a horse, cinematic lighting\"\n  }'

\n\n

Response

{\n  \"prompt\": \"An astronaut riding a horse, cinematic lighting\",\n  \"model\": \"Tongyi-MAI/Z-Image-Turbo\",\n  \"url\": \"https://jerrycans-file.hf.space/files/gen-1717000000000.png\",\n  \"content_type\": \"image/png\"\n}

\n\n

Want the image bytes directly (e.g. to pipe into a file)? Add ?raw=true:

curl \"$BASE/v1/image?prompt=a%20red%20bicycle&raw=true\" --output bike.png

\n\n \n

Calling it from code

\n\n

JavaScript

const base = \"$BASE\";\n\n// chat\nconst chat = await fetch(`${base}/v1/chat`, {\n  method: \"POST\",\n  headers: { \"Content-Type\": \"application/json\" },\n  body: JSON.stringify({ prompt: \"Write a haiku about the sea.\" }),\n}).then(r => r.json());\nconsole.log(chat.reply);\n\n// image\nconst img = await fetch(`${base}/v1/image`, {\n  method: \"POST\",\n  headers: { \"Content-Type\": \"application/json\" },\n  body: JSON.stringify({ prompt: \"A lighthouse at dusk\" }),\n}).then(r => r.json());\nconsole.log(img.url);

\n\n

Python

import requests\n\nbase = \"$BASE\"\n\nchat = requests.post(f\"{base}/v1/chat\", json={\n    \"prompt\": \"Explain gravity to a five year old.\"\n}).json()\nprint(chat[\"reply\"])\n\nimg = requests.post(f\"{base}/v1/image\", json={\n    \"prompt\": \"A fox in a snowy forest\"\n}).json()\nprint(img[\"url\"])

\n\n \n

Errors

Failures come back as JSON with the matching HTTP status:

{\n  \"error\": \"Provide a 'prompt' describing the image\",\n  \"detail\": null\n}

\n \n \n \n \n \n \n \n

Status	Meaning
`400`	Bad input — missing prompt, malformed JSON, etc.
`404`	Route doesn't exist.
`405`	Wrong HTTP method for that route.
`429`	Rate limited upstream — slow down and retry.
`5xx`	Model or upload host had a problem. The `detail` field usually says what.

\n\n

Endpoints at a glance

\n \n \n \n \n \n \n

Route	Method	Does
`/`	GET	This page.
`/health`	GET	Liveness check.
`/v1/chat`	GET / POST	Text generation.
`/v1/image`	GET / POST	Image generation → hosted URL.

\n\n \n

\n\n\n\n\n\n",[163],"pulls","branches","forks","envVars",{"_176":153,"_178":62},"connections",{"_176":62,"_177":62},"count","unfilledCount","emptyCount",{"_43":162,"_33":163,"_31":164,"_134":-5,"_41":42,"_40":165,"_136":25,"_38":197,"_13":153,"_34":20,"_36":167,"_24":27,"_140":-5,"_185":198,"_28":199},{"_43":155,"_33":156,"_31":157,"_134":-5,"_41":42,"_40":158,"_136":25,"_38":194,"_13":138,"_34":35,"_36":160,"_24":27,"_140":-5,"_185":195,"_28":196},{"_43":151,"_33":29,"_31":32,"_134":-5,"_41":42,"_40":14,"_136":27,"_38":191,"_13":153,"_34":20,"_36":37,"_24":27,"_140":-5,"_185":192,"_28":193},{"_43":142,"_33":143,"_31":57,"_134":-5,"_41":42,"_40":144,"_136":25,"_38":188,"_13":146,"_34":147,"_36":148,"_24":27,"_140":149,"_185":189,"_28":190},{"_43":131,"_33":132,"_31":133,"_134":-5,"_41":42,"_40":135,"_136":25,"_38":184,"_13":138,"_34":35,"_36":139,"_24":27,"_140":-5,"_185":186,"_28":187},["D",1780524869040],"children",[],[132],["D",1780525038944],[],[143],["D",1780525107729],[],[29],["D",1780524842321],[],[156],["D",1780525070161],[],[163],["D",1780519334637],{"_96":95,"_104":100},{"_96":95,"_97":98,"_99":100,"_18":55,"_101":102,"_103":27},{"_13":85,"_33":86,"_43":42,"_87":44,"_88":-5,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":45,"_94":95},"ctx",{"_78":-5,"_123":237,"_238":239,"_240":241,"_242":27,"_243":27,"_244":27,"_245":25,"_246":-5,"_247":-7,"_248":-7,"_249":62,"_250":27,"_251":27,"_252":-5,"_253":27,"_254":27,"_255":27,"_256":-5},"ENV",{"_208":209,"_210":211,"_212":86,"_213":214,"_215":216,"_217":218,"_219":220,"_221":222,"_223":224,"_225":226,"_227":228,"_229":230,"_231":232,"_233":234,"_235":236},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","5015b8df46402d8bba4f485ecc37aec202b0e06b","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Diff · v22