# AI3D Studio (remote · ai3d.studio)

Generate 3D models from images or text with AI, then retexture or convert them to STL, 3MF or OBJ.

- Trust score: 88/100 (high trust)
- Change this week: +3
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-04

## Components

- remote · `ai3d.studio`: 88/100 (this document), [markdown](https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d.md), [page](https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d)

## Channel facts

- Endpoint: `https://ai3d.studio/mcp`
- Transports: `streamable-http`
- Auth: `required`
- Version: `1.0.0`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-11-04.

- **Endpoint Security**: 97/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token.
  - HTTPS is enforced; there's no plaintext access path.
  - The HSTS (Strict-Transport-Security) header is present.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
  - The authorisation server supports Client ID Metadata Documents, the current MCP client-registration mechanism.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 66/100
  - AI-judged instruction clarity (excellent).
  - Context-footprint check failed: tool/resource definitions use about 2486 tokens (~191/item across 13 items; 13 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 73/100
  - Stability check failed: schema churn in the 26 days we've observed: 2 tool removals, 0 breaking changes, 0 auth/transport breaks, 0 additions.
- **Tool Coverage**: 91/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 67% of tool parameters carry a description.
  - Structured output schemas are declared (100% of tools); any adoption earns full credit.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - We read all 13 captured tool definition(s), and no name or description among them implies an irreversible operation.
  - An AI judge read all 14 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a current MCP spec version (2026-07-28).

## Install

### How do I install the AI3D Studio MCP server?

AI3D Studio is a hosted endpoint at https://ai3d.studio/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add --transport http studio-ai3d-ai3d-studio 'https://ai3d.studio/mcp'
```

### Cursor

```json
{
  "mcpServers": {
    "studio-ai3d-ai3d-studio": {
      "url": "https://ai3d.studio/mcp"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "studio-ai3d-ai3d-studio": {
      "type": "http",
      "url": "https://ai3d.studio/mcp"
    }
  }
}
```

### Codex

```toml
[mcp_servers.studio-ai3d-ai3d-studio]
url = "https://ai3d.studio/mcp"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "studio-ai3d-ai3d-studio": {
      "type": "remote",
      "url": "https://ai3d.studio/mcp",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add studio-ai3d-ai3d-studio --url 'https://ai3d.studio/mcp' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  studio-ai3d-ai3d-studio:
    url: "https://ai3d.studio/mcp"
```

### Netclaw

```json
{
  "McpServers": {
    "studio-ai3d-ai3d-studio": {
      "Transport": "http",
      "Url": "https://ai3d.studio/mcp"
    }
  }
}
```

### Vellum

```bash
assistant mcp add studio-ai3d-ai3d-studio -t streamable-http -u 'https://ai3d.studio/mcp'
```

### Other

```json
{
  "mcpServers": {
    "studio-ai3d-ai3d-studio": {
      "type": "http",
      "url": "https://ai3d.studio/mcp"
    }
  }
}
```

The mcpServers block is a cross-client convention. Remote transports vary, so check your client's docs.

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-11-03 (score 88, +1)

No change was recorded against any check on this day. Stability & Change Management went from 66 to 69.

### 2026-11-01 (score 87, +1)

No change was recorded against any check on this day. Stability & Change Management went from 59 to 63.

### 2026-10-30 (score 86, +1)

No change was recorded against any check on this day. Stability & Change Management went from 52 to 56.

### 2026-10-28 (score 85, +1)

No change was recorded against any check on this day. Stability & Change Management went from 46 to 49.

### 2026-10-26 (score 84, +1)

No change was recorded against any check on this day. Stability & Change Management went from 39 to 43.

### 2026-10-23 (score 83, +1)

No change was recorded against any check on this day. Stability & Change Management went from 29 to 33.

### 2026-10-21 (score 82, +1)

No change was recorded against any check on this day. Stability & Change Management went from 22 to 26.

### 2026-10-20 (score 81, +3)

- [security improvement] Authorization: fail → pass
- [security] First check of Authorization: pass
- [security] The server rewrote its instructions, which are the text every model session reads
- [security] The server changed its declared name: aiapps-agent-mcp → studio.ai3d/ai3d
- [cosmetic] “generate” added an optional parameter “expectedCredits”
- [cosmetic] “generate_3d” added an optional parameter “expected_credits”

## MCP tools (13)

### `list_models` (~221 tokens)

List models and prices

List the models this account can generate with, grouped by what they make (3D, image, …): each with its id, the scenes (input modes such as text-to-3d or image-to-3d) it supports, and the credits it costs at default options. Call this first to choose a model id, then estimate_cost for an exact price. Set include_inputs to also get the option fields of each scene (names, allowed values, ranges). An account whose app has Image Maker also gets presets: ready-made picture conversions (figurine, style, pose, …) to run with generate by passing options.preset. batches lists the scenes start_batch can run over several Library pictures at once, with the models each may use.

Input parameters:

- `available_only` (boolean): Leave out models that are listed but cannot be started.
- `include_inputs` (boolean): Include the option fields of every scene (larger result).
- `media_type` (string): Only models of this kind, e.g. '3d' or 'image'.

Output parameters:

- `app_id` (string)
- `batches` (array): Scenes start_batch can run over several pictures; empty when the app offers none.
- `count` (integer)
- `models` (array)
- `presets` (array): Image presets this app offers; empty when it has none or media_type is not image.

### `estimate_cost` (~141 tokens)

Estimate credit cost

Price a generation before starting it: the credits this model would charge for these options, and whether the current balance covers it. Uses the same validation and price calculation as generate_3d and generate, so for a valid request the figure is exactly what is charged. Omit scene to price every scene of the model. Nothing is charged and nothing is started.

Input parameters:

- `model` (string, required): A model id from list_models.
- `options` (object): The options you intend to pass; they change the price.
- `prompt` (string)
- `scene` (string): A scene id from list_models, e.g. 'image-to-3d'.

Output parameters:

- `available` (boolean)
- `balance` (number): Credits the account holds now.
- `media_type` (string)
- `model` (string)
- `quotes` (array)

### `get_balance` (~74 tokens)

Check credit balance

Get the credits this account can spend now (and how many of them come from the subscription or are permanent), the plan name when there is one, and — when you are connected with an API key that has a monthly credit limit — that limit, what the key has used this month, and when it resets. Read-only.

Output parameters:

- `account` (object)
- `api_key_limit`: Only for an API key with a monthly credit limit.
- `credits` (object)
- `plan` (string|null): The active subscription plan; null on pay-as-you-go.

### `generate_3d` (~246 tokens)

Generate a 3D model

Start a 3D generation and return its job id at once. mode picks the input: text (prompt), image (options with the image URL), multiview (several view images), sketch, or model (edit an existing 3D model). Take the model id and its option names from list_models, and check the price with estimate_cost first: this spends credits. Pass a stable idempotency_key when retrying so a retry replays the same job instead of billing twice. Then call get_job to follow it.

Input parameters:

- `expected_credits` (integer): The credits estimate_cost answered for this request. If the price has moved since, nothing is started or charged and the error (price_changed) gives the new price.
- `idempotency_key` (string): Reuse the same value when retrying this exact request.
- `mode` (string, required): Which input the model is given.
- `model` (string, required): A 3D model id from list_models.
- `options` (object): Model-specific options; take the field names from list_models with include_inputs, and price them with estimate_cost first.
- `prompt` (string)

Output parameters:

- `credits_charged` (number): Credits taken from the balance.
- `deduped` (boolean): True when an earlier call with the same idempotency key already started this job: nothing was charged again.
- `job_id` (string): Pass to get_job to follow progress.
- `media_type` (string)
- `model` (string)
- `outputs` (integer): How many results this job makes.
- `scene` (string|null)
- `status` (string): pending and processing are running; success, partial, failed and canceled are final. 'partial' means some outputs finished and the rest were refunded.

### `generate` (~328 tokens)

Start a generation

Start an async AI generation of any media type and return its job id. The input shape depends on the model (see list_models); for 3D prefer generate_3d. To run an image preset (list_models returns them: figurine, style, pose, convert), send mediaType "image", scene "image-to-image", options.image (the picture, or the subject then a reference), options.preset (its id) and optionally options.preset_addons and options.preset_background; the server composes the prompt, and prompt is then only your extra instructions. Bills credits: check the price with estimate_cost first. Pass a stable idempotencyKey when retrying — a retry with the same key replays the original job instead of billing twice. Then call get_job.

Input parameters:

- `expectedCredits` (integer): The credits estimate_cost answered for this request. If the price has moved since, nothing is started or charged and the error (price_changed) gives the new price.
- `idempotencyKey` (string)
- `mediaType` (string, required): Generation family, e.g. 'image' | 'video' | '3d' | 'music' — see list_models.
- `model` (string, required)
- `options` (object): Model-specific options; derive from list_models (include_inputs). For an image preset: preset (its id), preset_addons, preset_background.
- `prompt` (string): The prompt; with options.preset, only your additional instructions.
- `scene` (string)
- `workflowType` (string)

Output parameters:

- `credits_charged` (number): Credits taken from the balance.
- `deduped` (boolean): True when an earlier call with the same idempotency key already started this job: nothing was charged again.
- `job_id` (string): Pass to get_job to follow progress.
- `media_type` (string)
- `model` (string)
- `outputs` (integer): How many results this job makes.
- `scene` (string|null)
- `status` (string): pending and processing are running; success, partial, failed and canceled are final. 'partial' means some outputs finished and the rest were refunded.

### `get_job` (~96 tokens)

Get job status and results

Get a job started by generate_3d or generate: its status, progress, credits charged and refunded, and — when it has finished — the result files with their public download URLs, including other formats where the model returned them. Poll about every 5–10 seconds while status is pending or processing; stop when is_terminal is true.

Input parameters:

- `job_id` (string, required): The job id returned by generate_3d or generate.

Output parameters:

- `completed_at` (string|null)
- `created_at` (string|null)
- `credits` (object)
- `error`
- `files` (array): Result files with public download URLs; empty until ready.
- `is_terminal` (boolean): True once the job is final and will not change again.
- `job_id` (string)
- `media_type` (string)
- `model` (string)
- `outputs` (object)
- `progress` (number): 0 to 1: finished outputs over expected outputs.
- `scene` (string|null)
- `status` (string): pending and processing are running; success, partial, failed and canceled are final. 'partial' means some outputs finished and the rest were refunded.

### `list_job_files` (~144 tokens)

List job download formats

List the files a finished job already has, one per format: the main result plus any alternate formats the model returned (for 3D models, for example glb with fbx or usdz), each with a public download URL. Pass format to find one format. This does not convert: if the format asked for is not there, available_formats says what is, and a model can be exported to other formats from the web app model viewer.

Input parameters:

- `format` (string): A file format to look for, e.g. 'glb' or 'stl'.
- `job_id` (string, required): The job id returned by generate_3d or generate.

Output parameters:

- `available_formats` (array)
- `files` (array)
- `job_id` (string)
- `note` (string|null)
- `requested_format` (string|null)
- `status` (string): pending and processing are running; success, partial, failed and canceled are final. 'partial' means some outputs finished and the rest were refunded.

### `start_batch` (~281 tokens)

Start a batch over Library pictures

Run the same model, scene and options once for each of several pictures in the account’s Library — for example image to 3D for twelve product photos. Take the scenes and models from list_models (`batches`), and each picture’s asset_id from get_job or search_library on a finished image job. Nothing is charged now: each run is charged as it starts and refunded if it fails, but the balance must cover the whole quote, and for a key or connection with a monthly limit the whole quote is held against it until each picture runs (then its real cost counts) or is cancelled, refused or refunded (then nothing does). Returns a batch id at once; follow it with get_batch. Pass idempotency_key on retries so a repeat returns the same batch instead of queueing another.

Input parameters:

- `expected_credits` (integer): The quote you showed the person; a different price queues nothing.
- `idempotency_key` (string)
- `items` (array, required): One entry per picture: a Library image’s asset_id.
- `model` (string, required)
- `notify_by_email` (boolean)
- `options` (object): Options every run shares, as for generate; the picture field is filled per item.
- `project_id` (string)
- `prompt` (string)
- `scene` (string, required)

Output parameters:

- `balance` (number): Credits available when it was accepted.
- `batch_id` (string)
- `charged_credits` (number): What its runs were charged so far.
- `counts` (object)
- `created_at` (string)
- `item_count` (integer)
- `items` (array)
- `model` (string)
- `per_item_credits` (number)
- `quoted_credits` (number): What the batch costs if every picture runs.
- `reused` (boolean): An earlier call with this idempotency_key started it.
- `scene` (string)
- `settled_at` (string|null)
- `status` (string)

### `list_batches` (~68 tokens)

List recent batches

The account’s batches of one scene from the last 24 hours, newest first, with each picture’s progress — and how many pictures its plan lets one batch hold (allowance; 0 means the plan includes no batches).

Input parameters:

- `limit` (integer)
- `scene` (string, required)

Output parameters:

- `allowance` (integer)
- `batches` (array)

### `get_batch` (~71 tokens)

Get a batch’s progress

The state of a batch and of each picture in it: queued, running, done, failed or cancelled, the job_id of its run once started (get_job gives the files), credits charged, and whether retry_batch can run it again.

Input parameters:

- `batch_id` (string, required): The batch id start_batch returned.

Output parameters:

- `batch_id` (string)
- `charged_credits` (number): What its runs were charged so far.
- `counts` (object)
- `created_at` (string)
- `item_count` (integer)
- `items` (array)
- `model` (string)
- `per_item_credits` (number)
- `quoted_credits` (number): What the batch costs if every picture runs.
- `scene` (string)
- `settled_at` (string|null)
- `status` (string)

### `retry_batch` (~85 tokens)

Retry a batch’s failed pictures

Queue the batch’s failed pictures (and any stopped for want of credits) again as new runs, or only the item_ids given. Charged as each starts, like a new batch of that many pictures: the balance must cover them and they count against a monthly limit.

Input parameters:

- `batch_id` (string, required): The batch id start_batch returned.
- `item_ids` (array)

Output parameters:

- `batch_id` (string)
- `charged_credits` (number): What its runs were charged so far.
- `counts` (object)
- `created_at` (string)
- `item_count` (integer)
- `items` (array)
- `model` (string)
- `per_item_credits` (number)
- `quoted_credits` (number): What the batch costs if every picture runs.
- `requeued` (integer): Pictures queued again.
- `scene` (string)
- `settled_at` (string|null)
- `status` (string)

### `search_library` (~154 tokens)

Search my library

Search the Library of past generations kept by this account, newest first: filter by words in the prompt or title and by kind (3d, image, …). Returns the job id, model, status and result file URLs of each run. Use next_cursor to page. Only runs of this account are ever searched. For the full detail of one run call get_job.

Input parameters:

- `cursor` (string): The next_cursor of the previous result.
- `limit` (integer): How many to return, 1 to 50.
- `media_type` (string): Only runs of this kind, e.g. '3d' or 'image'.
- `query` (string): Words to look for in the prompt or title.

Output parameters:

- `items` (array)
- `next_cursor` (string|null): Pass as cursor for the next page; null at the end.
- `total` (integer): How many runs match in all.

### `convert_model` (~427 tokens)

Convert, compress or repair a Library model

Convert one of the account’s Library 3D models (a GLB) to GLB, STL, 3MF, OBJ or PLY on the server, optionally simplifying it, compressing its geometry, scaling it to a real-world size, moving its origin or repairing it for printing (holes, flipped faces, debris; STL, 3MF, OBJ or PLY), and file the result in the Library as a new creation made from it. Free (no credits) and answered at once. STL and 3MF are millimetres and Z-up for slicers; OBJ and PLY are geometry only, in metres; GLB keeps materials and textures and can be meshopt-compressed. Take asset_id from get_job or search_library (a file of a finished 3D job). Models over 50 MB or 2,000,000 triangles, and Draco-compressed GLBs, are refused. The same conversion run twice returns the same creation.

Input parameters:

- `asset_id` (string, required): A GLB file of one of your Library models.
- `compression` (string): GLB only: meshopt (default, smallest) or none.
- `format` (string): The file to make; glb when omitted.
- `name` (string)
- `origin` (string): Move the origin to the bottom centre (default when resizing) or the centre.
- `project_id` (string)
- `repair` (boolean): Repair for printing first: weld, drop broken and duplicate faces and loose debris, fix flipped faces, fill holes. stl, 3mf, obj, ply only; up to 200,000 triangles (simplify first for more).
- `resize_height` (number): Scale the model to this height, in metres.
- `resize_longest_side` (number): Scale the model so its longest side is this long, in metres. Not with resize_height.
- `simplify`: Share of triangles to keep: 1 (default), 0.75, 0.5 or 0.25.

Output parameters:

- `asset_id` (string): Its file, for another convert_model.
- `content_type` (string)
- `created` (boolean): False when the same conversion was already in the Library: that creation is returned.
- `format` (string)
- `job_id` (string): The new Library creation: pass to get_job like any job.
- `repair` (object): What repair found and fixed, when repair was asked for.
- `size_bytes` (integer)
- `triangles_after` (integer)
- `triangles_before` (integer)
- `url` (string): Public https download URL of the file.

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d#diagnostics

## Score history

- 2026-11-04: 88
- 2026-11-03: 88
- 2026-11-02: 87
- 2026-11-01: 87
- 2026-10-31: 86
- 2026-10-30: 86
- 2026-10-29: 85
- 2026-10-28: 85
- 2026-10-27: 84
- 2026-10-26: 84
- 2026-10-25: 83
- 2026-10-24: 83
- 2026-10-23: 83
- 2026-10-22: 82
- 2026-10-21: 82
- 2026-10-20: 81
- 2026-10-19: 78
- 2026-10-18: 77
- 2026-10-17: 77
- 2026-10-16: 76
- 2026-10-15: 76
- 2026-10-14: 75
- 2026-10-13: 75
- 2026-10-12: 75
- 2026-10-10: 75

## Common questions

### What is the AI3D Studio MCP server?

AI3D Studio is an MCP server listed in the public MCP registry as studio.ai3d/ai3d-studio. Generate 3D models from images or text with AI, then retexture or convert them to STL, 3MF or OBJ. This page covers its hosted endpoint (https://ai3d.studio/mcp).

### Is the AI3D Studio MCP server safe to use?

AI3D Studio scores 88 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the AI3D Studio MCP server expose?

AI3D Studio exposes 13 tools: list_models, estimate_cost, get_balance, generate_3d, generate, and 8 more. Their descriptions and schemas cost roughly 2,336 tokens of context every time the server is loaded.

### Does the AI3D Studio MCP server require authentication?

Yes. AI3D Studio asked us for credentials when we connected, so you will need to authorise it in your MCP client before it can do anything.

### Is the AI3D Studio MCP server still maintained?

AI3D Studio is still listed as active in the MCP registry. We last reached this channel on 4 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

## Links

- Remote endpoint: https://ai3d.studio/mcp
- Website: https://ai3d.studio/developers/mcp
- Changelog RSS feed: https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d.xml
- Changelog JSON feed: https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d.json
- HTML version of this page: https://verifymcp.io/servers/studio-ai3d-ai3d-studio/ai3d