Hosted MCP servers

Use the TopxAI catalogue from inside Claude Code, Cursor, VS Code or any client that speaks the Model Context Protocol. The servers authenticate with the same API key as the API, bill the same prices, and add a small flat fee per tool call.

How it is billed

A tool call that reaches a model is billed exactly like the same request made through the API, plus the flat call fee of the server it went through.
A call the provider fails is not billed: neither the model price nor the call fee.
A video job the provider fails after accepting it is refunded in full.
Every reply reports the exact charge, and each call appears in the usage log with its call fee as a separate line.

Servers

TopxAI Models

Ask any text model in the catalogue from inside another model's conversation: a second opinion, a review, or a model the host does not have.

Endpoint https://ai.topxea.com/mcp/models · $0.001 per call

Tool	What it does
`list_models`	List the text models this API key can ask, with their USD prices per million tokens and the call fee.
`ask_model`	Send a prompt to one text model of the TopxAI catalogue and return its reply. Billed at the model's token price plus a $0.001 call fee; the reply reports the exact charge.
`judge`	Ask TypeSafe's Jev (a System One model) typed questions about a piece of text and get calibrated numbers back, not prose: a probability for a yes/no question (type noul), one option with a probability per option (type choice, with criteria), or a value on a scale with a probability per anchor (type score, with anchors). Billed at Jev's input-token price plus a $0.001 call fee.

Example: list_models

You say: Which models can I ask through TopxAI, and what do they cost?

Tool call:

{"arguments":{},"name":"list_models"}

Result:

{"call_fee_usd":0.001,"models":[{"input_usd_per_million_tokens":5,"model":"gpt-6-astra","output_usd_per_million_tokens":25,"route":"openai-shared","unit":"per_million_tokens"}]}

Example: ask_model

You say: Ask GPT-6 for a second opinion on this migration plan and list the risks it sees.

Tool call:

{"arguments":{"max_cost_usd":0.05,"max_output_tokens":1024,"model":"gpt-6-astra","prompt":"Review this migration plan for risks: …"},"name":"ask_model"}

Result:

{"call_fee_usd":0.001,"charged":true,"charged_usd":0.01026,"finish_reason":"stop","model":"gpt-6-astra","request_id":"2026092208…","route":"openai-shared","usage":{"completion_tokens":288,"prompt_tokens":412}}

Example: judge

You say: Use Jev to tell whether this support ticket is urgent and what it is about.

Tool call:

{"arguments":{"questions":{"topic":{"criteria":{"account_access":"Login, password, permissions","billing":"Charges, invoices, refunds","integration":"Connecting a payment provider or an API","other":null},"instructions":"What is the message about?","type":"choice"},"urgency":{"instructions":"Does this message express urgency?","type":"noul"}},"state":"Hi, I have been trying to connect my Stripe account for 3 days and it keeps failing. I am losing sales. Please help ASAP."},"name":"judge"}

Result:

{"answers":{"topic":{"choice":"integration","confidence":1,"probabilities":{"account_access":0,"billing":0,"integration":1,"other":0},"type":"choice"},"urgency":{"noul":0.98,"type":"noul"}},"call_fee_usd":0.001,"charged":true,"charged_usd":0.00102,"model":"jev-1.13.0","request_id":"2026092208…","route":"typesafe-official","usage":{"input_tokens":447}}

TopxAI Images

Generate images with the catalogue's image model and receive them inline.

Endpoint https://ai.topxea.com/mcp/images · $0.01 per call

Tool	What it does
`generate_image`	Generate one or more images from a prompt with the TopxAI image model. Billed at the model's per-request price plus a $0.01 call fee. Returns the images inline and as signed links valid for one hour.

Example: generate_image

You say: Draw a poster of a fox in the snow, landscape.

Tool call:

{"arguments":{"n":1,"prompt":"A fox in the snow, poster style","size":"1536x1024"},"name":"generate_image"}

Result:

{"call_fee_usd":0.01,"charged":true,"charged_usd":0.06,"images":1,"model":"gpt-image-2.5-sunburst","n":1,"request_id":"2026092208…","size":"1536x1024","urls":["https://ai.topxea.com/v1/images/content/…"]}

TopxAI Videos

Submit a video generation job and poll it until the clip is ready.

Endpoint https://ai.topxea.com/mcp/videos · $0.01 per call

Tool	What it does
`generate_video`	Submit a text-to-video job. Billed per second of the requested length plus a $0.01 call fee when the provider accepts the job; a job that fails is refunded. Poll video_status with the returned request_id.
`video_status`	Report the status of a video job submitted with generate_video. Free of charge. When status is done, content_url serves the clip to the same API key.

Example: generate_video

You say: Make an 8-second 720p clip of waves at dusk.

Tool call:

{"arguments":{"duration":8,"prompt":"Waves at dusk, slow motion","resolution":"720p"},"name":"generate_video"}

Result:

{"billing_request_id":"2026092208…","call_fee_usd":0.01,"charged":true,"charged_usd":0.81,"duration":8,"model":"grok-imagine-video-1.5","request_id":"vid_8f3…","resolution":"720p","status":"pending"}

Example: video_status

You say: Is the video ready yet?

Tool call:

{"arguments":{"request_id":"vid_8f3…"},"name":"video_status"}

Result:

{"content_url":"https://ai.topxea.com/v1/videos/vid_8f3…/content","progress":100,"request_id":"vid_8f3…","status":"done"}

Connect

Every client uses the server URL with the API key in the Authorization header. With Claude Code:

claude mcp add --transport http topxai-models https://ai.topxea.com/mcp/models --header "Authorization: Bearer sk-your-key"

Cursor reads the same server from .cursor/mcp.json and VS Code from .vscode/mcp.json; the application page fills the snippets in with one of your own keys.

Get an API key · Every model's price · Docs · TopxAI

TopxAI