Claude connector · training

Your AI studio,
one sentence away.

Connect QynTro to Claude once. Then images, video, music, voiceovers, talking avatars and full PowerPoint decks answer to plain English — spending your credits, saved to your library.

https://qyntroai.com/mcp

This address is your connector. Paste it into Claude → Settings → Connectors — full walkthrough below.

Claude · connected to QynTro
FLUX Pro · Realism · Dev · Schnell GPT Image 2 · 1.5 · Mini Nano Banana Pro · Flash Stability Ultra · Core · SD 3.5 Kling 3 Pro · o3 · 2.1 · 1.6 Veo 3.1 · 3.1 Fast · 3 Seedance · Luma · MiniMax ElevenLabs Music · Voice HeyGen Avatars PPTX Presentations

Connect QynTro to Claude

Five clicks, about two minutes, on claude.ai or the Claude desktop app. You'll need a QynTro account (free works) and a paid Claude plan — custom connectors are a paid-plan feature on Claude's side.

1

Open Claude's Settings

On claude.ai, click your initials in the bottom-left corner, then choose Settings.

Claude profile menu with Settings highlighted
2

Go to Connectors

In the Settings sidebar, under Customize, click Connectors.

Claude settings sidebar with Connectors highlighted
3

Click Add → Add custom connector

Top-right of the Connectors page: open the Add menu and pick Add custom connector.

Connectors page with the Add menu open and Add custom connector highlighted
4

Name it QynTro and paste the address

Name: QynTro  ·  Remote MCP server URL: https://qyntroai.com/mcp — leave the Advanced settings empty and click Add.

Add custom connector dialog with the QynTro MCP address filled in
5

Click Connect, sign in, press Allow

Click Connect on the new QynTro card — a QynTro page opens. Sign in with your QynTro account if asked, then press Allow. That's it: you're live.

QynTro authorization screen with the Allow button highlighted

Tip: be signed in at qyntroai.com in the same browser first — then step 5 is a single click. You can disconnect any time from the same Connectors screen.

Just ask — say “QynTro”

Plain sentences work — no commands, no syntax. Mention QynTro by name in the prompt so Claude knows to use your connector instead of answering by itself:

Using QynTro, create an image of a snow leopard on a mountain at dawn
Here's your image — 10 credits, saved to your QynTro library.
Use QynTro to make a vertical video of waves crashing on rocks at sunset
Rendering a 5-second clip on the low-cost video model — job 214, I'll have the link in about two minutes.

Built-in cost safety: when you don't name a model or length, video defaults to a short 5-second clip on the cheapest capable model, and audio / 4K stay off unless you ask. You never burn credits by accident.

AI Images — 14 models by name

Name any model — casually or precisely — and that exact model runs. Sizes work the same way: “make it 16:9”, “portrait”, “square”. Up to 4 images per request.

On QynTro, create an image using GPT Image 2 — a vintage travel poster for Mars
Use QynTro's Nano Banana Pro for a product shot of a perfume bottle
Using QynTro, make 4 logo drafts with FLUX Schnell — a fox coffee brand

Image models

ask “list the QynTro models” for live prices & sizes
Say thisModelBest forCost
flux/schnellFLUX SchnellInstant drafts — 1 creditcheapest
flux/devFLUX DevBalanced qualitylow
flux-realismFLUX RealismPhotoreal peoplemedium
flux-pro/newFLUX Pro defaultBest all-roundmedium
gpt-image-2GPT Image 2Text in images, precise promptsmedium
gpt-image-1.5GPT Image 1.5Strong OpenAI qualitymedium
gpt-image-1-miniGPT Image MiniCheap OpenAI optionlow
gemini-2.5-flash-imageNano Banana FlashFast all-rounder · square onlylow
gemini-3-pro-image-previewNano Banana ProHighest detail · square onlypremium
core  /  ultraStability Core / UltraArt styles, 8 aspect ratioslow medium
sd3.5-largeSD 3.5 LargeIllustrationmedium
sd3.5-large-turboSD 3.5 Large TurboFast illustrationlow
sd3.5-mediumSD 3.5 MediumBudget illustrationlow

FLUX takes 6 aspect presets (square, 4:3, 16:9, portrait & landscape) · GPT Image takes square, 1536×1024 and 1024×1536 · Nano Banana is square-only · Stability takes 8 ratios from 9:21 to 21:9.

AI Video — charged per second

Say nothing about model or length and QynTro protects your wallet: 5 seconds on the cheapest capable model. Name a model and a length, and you get exactly that.

Use QynTro to make a 5 second video of a paper boat floating down a rainy street
On QynTro, generate a 10 second video with Kling 3 Pro of a fox running through snow
Using QynTro, make an 8 second Veo 3.1 clip with audio — city traffic timelapse

Video models · text to video

price = per-second rate × length
Say thisModelLengthsCost
kling-videoKling 1.6 Pro default5s · 10scheapest
kling-o3Kling o3 Pro5s · 10smedium
kling-3-proKling 3 Pro5s · 10smedium
google-veo-31-fastVeo 3.1 Fast4s · 6s · 8smedium
seedance-2-fastSeedance 2 Fast5s · 10shigh
kling-video-21-masterKling 2.1 Master5s · 10shigh
google-veo-31Veo 3.1 · audio option4s · 6s · 8spremium
google-veo3Veo 38s onlypremium
minimax-videoMiniMax · Mochi · Hunyuanfixed cliplow
luma-dream-machineLuma Dream Machinefixed clip · 6 ratiosmedium

Audio and higher resolution on the Veo models multiply the price up to ~3.5× — leave them off unless the shot needs them.

Video takes 1–4 minutes, so it renders in the background while you keep chatting:

▶

You ask

“Use QynTro to make a 5 second video of ocean waves at dusk”

⏳

QynTro renders

You get a job id and the credit cost immediately

✓

You collect

Ask “is my QynTro video ready?” — the finished link comes back

Failed renders refund themselves. You're only ever charged for video that actually arrives — the clip also waits in your QynTro library.

Image → Video — animate any picture

Turn a still image into a moving clip. It chains naturally: generate an image first, then animate the one you like — Claude passes the picture along for you.

Using QynTro, create an image of a neon-lit alley in the rain — cyberpunk style
Perfect. Now animate that image into a 5 second video, slow push-in
Animating with Kling 1.6 Pro — job i18, about two minutes. Refunded automatically if it fails.

You can also paste any public image URL. Same cost rule as video: no model named = cheapest model, 5 seconds. Extra models here: kling-video-21-standard, kling-video-21-pro, seedance-2-5 (480p/720p) and google-veo2.

Image requirements: a public https link to a PNG, JPEG or WEBP under 15 MB. Images you just generated with QynTro always qualify.

AI Music — original tracks

Describe the track — genre, mood, tempo, instruments — and ElevenLabs Music composes an original piece. 10 to 120 seconds, default 30. Vocals by default; say “instrumental” to skip them.

Use QynTro to make a 30 second upbeat lo-fi track for a product demo, instrumental
Composing… done — 2 credits. Here's your MP3, and it's saved in your QynTro library.
Using QynTro, create a 90 second cinematic orchestral piece — rising tension, big drums
Done — 5 credits for the longer bracket. Download link below.

Tracks up to 30 seconds cost 2 media credits; 31–120 seconds cost 5. The call waits for the finished audio (30–90 seconds) — nothing is charged if composing fails. Need longer than 2 minutes? The QynTro dashboard goes up to 10.

AI Voiceover — text to speech

Turn any text into a natural MP3 voiceover with ElevenLabs and OpenAI voices. Ask for a gender and language and QynTro picks a matching voice — or name an exact voice from the list.

Using QynTro, read this in an English male voice: “Welcome to the future of creativity.”
Done — voiced by Adam (ElevenLabs), 44 character credits. MP3 below.
What QynTro voices do you have in Spanish?
Here are the voices — name, vendor, gender and language for each. Pick one and I'll use it.

Different balance: voiceover spends character credits — 1 per character of text — a separate balance from your media credits. Check it any time in the QynTro dashboard.

One voice per request, MP3 output. The spoken language follows your text — the language you ask for is used to pick a matching voice.

AI Avatars — a presenter on demand

A studio-quality HeyGen avatar speaks your script to camera — perfect for promos, explainers and social clips. Start by seeing who's available:

List my QynTro avatars
You have Amelia (female, business), Marcus (male, casual), Yuki (female, studio)… pick one and give me a script.
Make an avatar video with Amelia saying: “Welcome to QynTro — every AI model, one subscription.” Portrait, please.
Rendering — job a12, 9:16 portrait, 12 credits for a ~4 second script. Ask me “is a12 ready?” in a few minutes.
script length ÷ 15 ≈ seconds × 3 credits / second = total cost

A 300-character script ≈ 20 seconds ≈ 60 credits. Scripts up to 1,500 characters. Landscape (16:9) or portrait (9:16). Ask “list QynTro avatar voices” to choose the speaking voice — otherwise a clear English voice is picked for you. Failed renders refund automatically.

Presentations — a real .pptx

Give QynTro a topic and get back a finished PowerPoint file — researched content, themed design, download link in about a minute. The deck also stays editable in your QynTro dashboard.

Use QynTro to create a 6-slide presentation on coffee culture, midnight theme
Here's your .pptx — 12 credits (2 per slide). It's also in your dashboard if you want to edit slides.
Using QynTro, make a detailed 5-slide investor deck about our fitness app, with AI images, corporate theme
Done — 5 slides with generated images, 50 credits. Download link below.

Themes — dark and light, default highlighted:

carbon-lime midnightobsidiandeep-ocean aubergineforestcrimsonslate paperivorymistsandmintblushcorporate

Pricing: 2 credits per slide, +8 per slide when AI images are on. Default 6 slides, Medium detail, English, no images. Up to 20 slides text-only, 6 with images. Detail levels: Brief · Medium · Detailed · Extensive — and any language: “…in Hindi”, “…in Spanish”.

Credits & jobs, always in view

Images, video, music, avatars and presentations spend the media credits on your QynTro plan; voiceover spends character credits. Same balances as the dashboard — every reply states what it cost, so nothing is a surprise.

How many QynTro credits do I have left?
You have 1,240 media credits — about 124 FLUX Pro images, or ~12 short videos, or a stack of decks.
Are my QynTro jobs done?
Video 214 is READY — link below. Animation i18 is still rendering (2 min so far). Avatar a12 is READY.

Job ids: plain numbers are text-to-video, i… is image-to-video, a… is an avatar video. Anything that fails refunds itself automatically.

Get better results

Say “QynTro” in the prompt

“Using QynTro, …” routes the request to your connector every time — especially if you have other connectors installed.

Describe the shot, not just the subject

Light, mood, lens, camera move. “Slow push-in, golden hour, 85mm” beats “a dog”.

Draft cheap, finish premium

Explore with flux/schnell at 1 credit, then re-run the winner on flux-pro/new or Nano Banana Pro.

Chain image → video

“Create an image of a neon alley — then animate it into a 5 second video.” Claude passes the image along automatically.

Short scripts, cheap avatars

Avatar cost follows script length — 15 characters ≈ 1 second ≈ 3 credits. Tighten the script before rendering.

Iterate in the chat

“Same but at night”, “make it vertical”, “more cinematic” — it re-generates with your change.

Troubleshooting

Claude answers by itself instead of using QynTro
Mention QynTro by name — “Using QynTro, create…” — and check the connector shows as connected under Settings → Connectors. In a chat you can also open the tools menu (the sliders icon) and make sure QynTro is enabled for that conversation.
“Not enough media credits”
Top up or upgrade in your QynTro dashboard — or ask for a cheaper route: flux/schnell for images, kling-video for video, fewer slides, or a shorter avatar script.
The connector won't connect
Sign in at qyntroai.com in the same browser, then remove and re-add the connector. Custom connectors also need a paid plan on Claude's side.
My video / avatar is still rendering
Most clips finish in 1–4 minutes — just ask again (“is my QynTro video ready?”). If a render fails, the credits come back automatically.
A model name isn't recognised
Ask “list the QynTro models” for the exact live names and prices, then use one from the list.
“Rate limit” message
Generations are capped at 20 per minute per account across all tools. Wait a moment and continue — nothing was charged for the blocked call.
What about ChatGPT?
A ChatGPT app is on our roadmap. Today the QynTro connector is built and tested for Claude — web and desktop.