> **Reading as an AI agent?** This is the Markdown version of https://docs.meshy.ai/api/changelog.
>
> - Full docs index: https://docs.meshy.ai/llms.txt
> - Single-fetch full content: https://docs.meshy.ai/llms-full.txt
> - Tool-calling access via MCP server: https://docs.meshy.ai/api/ai

---
# Changelog

See all of the latest features and updates to the Meshy API and plugins.

Prefer a live feed? The latest entries also show up on the [Developer Platform](https://www.meshy.ai/developers) overview page.

---

## October 2026
### `Oct 6`

- [Analyze Printability](/api/analyze-printability#create-an-analyze-printability-task) and [Repair Printability](/api/repair-printability#create-a-repair-printability-task) can now check wall thickness at the size you plan to print. Send `printer_type`, a print size, or both: analyze reports thin walls in `printability.thin_walls`, and repair thickens them when needed, and can reuse the printer and size of an analyze task passed as `input_task_id`. Integrations that don't use these fields behave as before, and pricing is unchanged.

## September 2026
### `Sep 29`

- The legacy `/v2/text-to-3d` path (without the `/openapi` prefix) is deprecated. It keeps working unchanged for now, but will be removed in a future release — switch to [`/openapi/v2/text-to-3d`](/api/text-to-3d), which takes the same request and returns the same response. Responses on the legacy path now carry a `Deprecation` header and a `Link` header pointing to the new path.

### `Sep 28`

- API responses now include [rate limit headers](/api/rate-limits#rate-limit-headers), including `Retry-After` to tell you how long to wait before retrying or polling again.
- [Multi-Color Print](/api/multi-color-print) accepts a new `printer_brand` parameter that writes a matching slicer preset into the 3MF, so the file opens ready to print on Anycubic, Creality, Elegoo, Flashforge, Prusa, Qidi, Snapmaker and AnkerMake printers as well as Bambu Lab (the default). The page now also documents `max_depth`.

### `Sep 18`

- Added `meshy-7.1` as an `ai_model` value for [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task),
  [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) and
  [Text to 3D](/api/text-to-3d#create-a-text-to-3d-preview-task): same parameters and credit cost as
  `meshy-7`, higher geometry detail. `latest` now resolves to `meshy-7.1`; `meshy-7` is deprecated but keeps working unchanged.
- Added `geometry_resolution` (`standard` / `2k` / `4k`) to the same APIs, replacing the now-deprecated
  `ultra_mode` (`ultra_mode: true` still works and equals `"2k"`). `4k` requires Meshy 7.1 (`meshy-7.1` or `latest`), is not
  available on Multi-Image to 3D, and costs the same 5 extra credits as `2k`.

### `Sep 16`

- The [Auto Split API](/api/auto-split) now accepts a textured `input_task_id`. It previously rejected one with `400`, which meant generating the same mesh a second time with `should_texture: false` just to split it. Note what a split returns: it cuts geometry and rebuilds each part's surfaces, so the input's texture is not carried over and the result is an untextured model with one flat color per part — that was already true of every split and is unchanged.
- `meshy-5` will be retired on **October 10, 2026**. It is no longer documented as an `ai_model` value for [Text to 3D](/api/text-to-3d#create-a-text-to-3d-preview-task), [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) or [Retexture](/api/retexture#create-a-retexture-task), and after October 10 requests that send `ai_model: "meshy-5"` will return an error.
- The shared **test-mode API key** has been retired; requests that still send it now return `401 Unauthorized`. If you copied it into code or CI, replace it with your own key from the [API Keys page](https://www.meshy.ai/developers/keys). That key always returned the same sample model regardless of input, so it never showed what a request would actually generate. To see what the API returns before building against it, use the [Playground](/api/developer-platform#playground) in the Developer Platform, which runs a real task in your browser. On plans without API access it runs Image to 3D without an API key and uses your account's credits like any other task.

### `Sep 15`

- Added `gpt-image-2-5-flare` and `gpt-image-2-5-sunburst` to the documented `ai_model` values for the [Text to Image](/api/text-to-image#create-a-text-to-image-task) and [Image to Image](/api/image-to-image#create-an-image-to-image-task) APIs. These are OpenAI GPT Image 2.5 tiers, priced the same as `gpt-image-2` (Text to Image: 9 credits; Image to Image: 12 credits).
- The GPT Image models (`gpt-image-2`, `gpt-image-2-5-flare`, `gpt-image-2-5-sunburst`) now also accept the `4:3` and `3:4` aspect ratios on the [Text to Image](/api/text-to-image#create-a-text-to-image-task) and [Image to Image](/api/image-to-image#create-an-image-to-image-task) APIs, in addition to `1:1`, `16:9`, `9:16`, `3:2`, and `2:3`.

### `Sep 14`

- The [Image to Image API](/api/image-to-image#create-an-image-to-image-task) now accepts `input_task_id` — chain directly from a completed Text to Image or Image to Image task instead of copying its output URLs into `reference_image_urls`.

### `Sep 11`

- Behavior change: `DELETE` on a task that is already `IN_PROGRESS` is now rejected with `409 Conflict`, and the task is left running. It previously returned `200 OK`, stopped the worker and discarded the result — while keeping the credits, because only `PENDING` tasks are refundable, so the call cost you both the model and the credits with nothing in the response to say so. If your integration deletes tasks in order to cancel them, note that a task the worker has already started cannot be cancelled: wait for `SUCCEEDED`, `FAILED` or `CANCELED`, then delete it. `PENDING` tasks are still deleted and refunded exactly as before, and terminal tasks are still deleted. Applies to every task family under `/openapi`. See [Errors](/api/errors).
- Added `meshy-6-lite` to the documented `ai_model` values for the [Text to 3D](/api/text-to-3d#create-a-text-to-3d-preview-task), [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) and [Retexture](/api/retexture#create-a-retexture-task) APIs. It is the successor to `meshy-5`: same parameters, same credit cost (5 for the mesh, 15 with texture, 10 for a retexture).
- `meshy-5` is now deprecated across every API that accepts it. It keeps working and nothing about your request or your bill changes, but it will stop accepting new traffic in a future release — if your integration sends `ai_model: "meshy-5"`, migrate to `meshy-6-lite`.

### `Sep 9`

- New [Auto Split API](/api/auto-split): Split an untextured Meshy 6 or Meshy 7 model into separately printable parts — automatically, by the parts you name (`by_parts`), or by color region (`by_color`) — with optional mortise-and-tenon connectors and thin-wall reinforcement. Exports every part as its own object in `glb`, `obj`, `fbx`, `usdz`, `blend`, or `3mf`, either assembled or laid out on the build plate (`layout`). 10 credits per task.

### `Sep 7`

- Added the `name_text` parameter to the [Creative Lab Keychain API](/api/creative-lab-keychain#create-a-keychain-prototype-task) prototype stage: engrave a name of up to 10 characters on the keychain, the same personalization the web app offers. The build stage inherits the engraving automatically, and the prototype task object now returns `name_text` so you can confirm the exact text before building. Note that `name` remains a display-only task label and is not engraved.

### `Sep 5`

- Added the [Creative Lab Fidget Pixel API](/api/creative-lab-fidget-pixel): turn a photo into a multi-color 3D-printable pixel-art fidget board in two stages. `POST /prototype` (6 credits) pixelizes the photo into a pixel-art image — pass `type` as `person` or `other` to pick the pixelization style — and `POST /build` (30 credits) samples it onto a `16` or `32` grid and turns every pixel into an interlocking `square` or `hex` piece, delivered as a single color-tagged 3MF for multi-filament printing. Piece geometry is tunable via `piece_size_mm`, `color_count`, and `piece_height_mm`.

### `Sep 4`

- Added [`GET /openapi/v1/animations/library`](/api/animation-library#list-animations) to the [Animation Library](/api/animation-library): fetch every preset action the [Animation API](/api/animation#create-an-animation-task) accepts — `action_id`, name, category, and an animated GIF preview — with search and category filters, so you can build your own action picker instead of reading the reference table. Free; consumes no credits.
- Added the [Creative Lab Collapsible Fidget API](/api/creative-lab-fidget-collapsible): turn a photo into a print-in-place collapsible fidget toy, where the subject's silhouette becomes nested concentric rings that collapse flat and stretch out again, printed in one piece with no assembly. A single `POST` carries the image through to the finished model — there is no prototype/build stage pair and no geometry parameters to tune. Each task costs 6 credits and returns both a printable STL and a GLB for previewing.

### `Sep 3`

- The [Creative Lab Lamp API](/api/creative-lab-lamp) prototype stage now returns a 3D model: `POST /prototype` turns the concept image into a hollow lampshade GLB, and the prototype task object gains `model_urls.glb` and `thumbnail_url` alongside `image_urls`. The build stage runs only the lamp processor on that model and now produces an open-bottom lampshade with a base plate for the light-source fixture. Pricing moves with the work: prototype is 30 credits and build 6 credits (previously 6 and 30). The prototype `text` parameter was removed — `image_url` is required.

### `Sep 2`

- The [Animation API](/api/animation#create-an-animation-task) now accepts `action_ids` — apply up to 10 preset actions to a rigged character in one request and receive a **single file containing one animation clip per action**, ready to drive from a state machine in a game engine. Clips keep the order you list them in and carry the animation's library name. Costs 3 credits per action, and `post_process` applies to the merged file. See the [Animation Library Reference](/api/animation-library) for available ids and the [pricing table](/api/pricing).

### `Sep 1`

- Added the `ultra_mode` parameter to the [Multi-Image to 3D API](/api/multi-image-to-3d#create-a-multi-image-to-3d-task): enables Ultra generation on `ai_model` `meshy-7` (or `latest`) for higher-fidelity geometry with finer surface detail, for 5 extra credits — the same Ultra pass the [Image to 3D API](/api/image-to-3d#create-an-image-to-3d-task) offers, now available with multi-view inputs.
- Removed `meshy-t1` from the documented `ai_model` values for the [Image to 3D API](/api/image-to-3d#create-an-image-to-3d-task)'s Smart Topology `model_type`. Smart Topology now only documents `meshy-t2`; new integrations should use it instead. Meshy T1 is planned for deprecation — if your integration currently sends `ai_model: "meshy-t1"`, migrate to `meshy-t2`.

## August 2026
### `Aug 28`

- Added the [Usage API](/api/usage#list-usage-records) for Studio and Enterprise teams: programmatically export historic API task usage — endpoint, status, timestamps, the API key that ran each task, and `consumed_credits` — with standard pagination and time-window, endpoint, and status filters, for cost tracking and finance systems.

### `Aug 26`

- The [Animation API](/api/animation#create-an-animation-task) now accepts `motion_task_id` — apply a motion clip generated with the [Text to Motion API](/api/text-to-motion) onto your rigged biped character, as an alternative to a preset `action_id`. Same 3 credits per task.

### `Aug 24`

- Expanded `aspect_ratio` support for `gpt-image-2` on the [Text to Image](/api/text-to-image#create-a-text-to-image-task) and [Image to Image](/api/image-to-image#create-an-image-to-image-task) APIs to add `16:9` and `9:16`. `gpt-image-2` now accepts `1:1`, `16:9`, `9:16`, `3:2`, and `2:3`; `4:3` and `3:4` remain unsupported for this model.

### `Aug 21`

- Added the [Text to Motion API](/api/text-to-motion): generate a character motion clip from a natural-language description, no character model required. Choose `prime` mode (10 credits, FBX output) for the highest quality or `swift` mode (3 credits, BVH output) for fast, economical generation, with clip durations of 2–10 seconds.

### `Aug 20`

- Added the `style` parameter to the [Multi-Color Print API](/api/multi-color-print#create-a-multi-color-3d-print-task): choose `realistic` (default) for texture-accurate colors, or `cartoon` for flat, uniform color regions and a smaller 3MF.
- Added the `remove_background` parameter to every Creative Lab prototype endpoint — [Figure](/api/creative-lab-figure#create-a-figure-prototype-task), [Vinyl Figure](/api/creative-lab-vinyl-figure#create-a-vinyl-figure-prototype-task), [Brick Figure](/api/creative-lab-brick-figure#create-a-brick-figure-prototype-task), [Keycap](/api/creative-lab-keycap#create-a-keycap-prototype-task), [Keychain](/api/creative-lab-keychain#create-a-keychain-prototype-task), [Fridge Magnet](/api/creative-lab-fridge-magnet#create-a-fridge-magnet-prototype-task) and [Lamp](/api/creative-lab-lamp#create-a-lamp-prototype-task). Set `remove_background: true` to receive the prototype image as a transparent RGBA PNG with the background removed.

### `Aug 18`

- Added the `texture_image_urls` parameter to the [Multi-Image to 3D API](/api/multi-image-to-3d#create-a-multi-image-to-3d-task): provide 1–4 images of the same object to guide texturing, instead of the single `texture_image_url`. Requires `ai_model` `meshy-7` or `latest`, and is independent of `image_urls` — the two lists can contain different images and different numbers of images.

### `Aug 17`

- Added the `remove_background` parameter to the [Text to Image API](/api/text-to-image#create-a-text-to-image-task) and [Image to Image API](/api/image-to-image#create-an-image-to-image-task). Set `remove_background: true` to receive the generated image as a transparent RGBA PNG with the background removed.

### `Aug 13`

- Added `ai_model: "meshy-7"` to the [Text to 3D API](/api/text-to-3d#create-a-text-to-3d-preview-task): our newest 3D generation model, with higher-fidelity geometry than Meshy 6. It also introduces the `ultra_mode` parameter on the preview task, which enables Ultra generation for even finer surface detail for 5 extra credits. To texture with Meshy 7, pass `ai_model: "meshy-7"` on the refine task or omit `ai_model` entirely.
- Added the **Smart Topology** `model_type` to the [Text to 3D API](/api/text-to-3d#create-a-text-to-3d-preview-task). Set `model_type: "smart-topology"` to generate with `meshy-t2` — cleaner topology, natively separated parts, triangle output, and a face count you can set with `target_polycount` (100 to 15,000, default 4,000). A Smart Topology preview costs 5 credits instead of 20; see [Pricing](/api/pricing). The existing `model_type: "lowpoly"` is now deprecated but continues to serve traffic for backward compatibility.

### `Aug 12`

- Added `ai_model: "meshy-7"` to the [Image to 3D API](/api/image-to-3d#create-an-image-to-3d-task): our newest 3D generation model, with higher-fidelity geometry than Meshy 6. It also introduces the `ultra_mode` parameter, which enables Ultra generation for even finer surface detail.
- Added `ai_model: "meshy-7"` to the [Multi-Image to 3D API](/api/multi-image-to-3d#create-a-multi-image-to-3d-task): Meshy's multi-view generation model, which conditions on all of your input views together and drives texturing from them.
- Added `ai_model: "meshy-7"` to the [Retexture API](/api/retexture#create-a-retexture-task): our newest texturing model, with sharper surface detail and cleaner PBR materials than Meshy 6. It also introduces the `multiview_image_urls` parameter — provide 1–4 ordered views of the same object to drive multi-view texturing.

### `Aug 4`
- Expanded `model_url` input support to `.glb`, `.gltf`, `.obj`, `.fbx`, `.stl` for the [Analyze Printability](/api/analyze-printability) and [Repair Printability](/api/repair-printability) endpoints (previously `.glb`, `.stl`, `.obj`). Repair returns the repaired asset in its input format.

## July 2026
### `Jul 27`

- New **[Creative Lab — Brick Figure API](/api/creative-lab-brick-figure)**: turn a photo into a brick-style collectible 3D minifigure. Same two-stage workflow as the [Figure API](/api/creative-lab-figure) — `prototype` (6 credits) generates a styled concept image, `build` (30 credits) turns it into a textured 3D model with GLB and OBJ + MTL outputs. See [Pricing](/api/pricing) for the full breakdown.

### `Jul 22`

- New **[Creative Lab — Keycap API](/api/creative-lab-keycap)**: turn a photo into a full-color custom mechanical keyboard keycap. Two-stage workflow — `prototype` (12 credits) generates a finished-keycap design render, then `build` (50 credits) turns the design you pick (via `candidate_id`) into a textured 3D keycap model with GLB and OBJ bundle outputs, exported at real-world millimeter scale. Currently supports the standard Cherry MX 1u base (`cherry-mx-1x1-r1`); more standard sizes are planned. See [Pricing](/api/pricing) for the full breakdown.
### `Jul 21`

- We now support **8K base color textures** via a new `texture_resolution` parameter (`2k` / `4k` / `8k`) on the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task), and [Retexture](/api/retexture#create-a-retexture-task) endpoints. At `8k`, PBR maps are generated at 4K and no emission map is produced. 8K texture generation costs 15 credits (`2k`/`4k` remain 10). See [Pricing](/api/pricing).
- The `hd_texture` boolean is deprecated in favor of `texture_resolution` — it is still honored and equivalent to `texture_resolution: "4k"` for backward compatibility.

### `Jul 20`

- New **[Creative Lab — Vinyl Figure API](/api/creative-lab-vinyl-figure)**: turn a photo into a big-head vinyl collectible 3D figure. Same two-stage workflow as the [Figure API](/api/creative-lab-figure) — `prototype` (6 credits) generates a styled concept image, `build` (30 credits) turns it into a textured 3D model with GLB and OBJ + MTL outputs. See [Pricing](/api/pricing) for the full breakdown.

### `Jul 13`

- Added a new **Smart Topology** `model_type` to the [Image to 3D](/api/image-to-3d) API, offering two models via `ai_model`: `meshy-t1` — the current low-poly model — and `meshy-t2` — a new model with cleaner topology, natively separated parts, triangle output, and a face count you can set with `target_polycount`. `meshy-t2` is the default under `smart-topology`. The existing `model_type: lowpoly` is now deprecated but continues to serve traffic for backward compatibility.

### `Jul 7`

- Added the `aspect_ratio` parameter to the [Image to Image](/api/image-to-image) API, so you can control the output image ratio when editing. Allowed values are `1:1`, `16:9`, `9:16`, `4:3`, `3:4` for the nano-banana models and `1:1`, `3:2`, `2:3` for `gpt-image-2`. Optional; defaults to `1:1`, and cannot be combined with `generate_multi_view`.

## June 2026
### `Jun 17`

- New **[API Playground](/api/playground)** docs page: an interactive console for trying endpoints without writing code, then copying the equivalent cURL, Python, or JavaScript. Available on the Pro, Premium, Ultra, Studio, and Enterprise plans.

### `Jun 15`

- New **[UV Unwrap API](/api/uv-unwrap)**: generate a high-quality UV layout for an existing 3D model — the prerequisite step before texturing, or any time you need a clean, non-overlapping UV layout for downstream tools. Takes either an `input_task_id` (a completed Image to 3D, Text to 3D, or Remesh task) or a `model_url` (`.glb` only) and returns a "UV white model" with brand-new UV coordinates. 5 credits per call. Supports meshes of up to 40,000 faces (run [Remesh](/api/remesh) first if larger); quad and n-gon meshes are triangulated during UV generation.

### `Jun 12`

- Updated **Creative Lab API** Build pricing to **30 credits** per call for [Keychain](/api/creative-lab-keychain), [Fridge Magnet](/api/creative-lab-fridge-magnet), and [Figure](/api/creative-lab-figure) (previously 20). [Lamp](/api/creative-lab-lamp) is unchanged at 30 credits. Prototype pricing is unchanged at 6 credits per image. See [Pricing](/api/pricing) for the full breakdown.

### `Jun 5`

- Added the optional `alpha_thumbnail` parameter to the [Image to 3D](/api/image-to-3d), [Multi-Image to 3D](/api/multi-image-to-3d), [Text to 3D](/api/text-to-3d), [Retexture](/api/retexture), [Remesh](/api/remesh), and [Repair Printability](/api/repair-printability) APIs. When set to `true`, the task additionally renders a transparent-background (RGBA) version of the preview and returns it under the new `alpha_thumbnail_url` response field. The existing `thumbnail_url` is unchanged. Defaults to `false`.

### `Jun 1`

- New **Creative Lab APIs**: a family of product-scoped endpoints that turn a source photo (or text prompt) into a 3D-printable consumer product. Four products at launch — [Keychain](/api/creative-lab-keychain), [Fridge Magnet](/api/creative-lab-fridge-magnet), [Figure](/api/creative-lab-figure), and [Lamp](/api/creative-lab-lamp). Each follows a two-stage `prototype` → `build` workflow chained via `input_task_id`, served under product-scoped URLs (`/openapi/creative-lab/<product>/v1/...`) with per-product versioning so products can evolve on independent version lines. Prototype: 6 credits per call. Build: 30 credits per call.

---

## May 2026
### `May 26`

- Added `multi_view_thumbnails` parameter to the [Multi-Image to 3D](/api/multi-image-to-3d) API. When set to `true`, the task additionally renders four cardinal-view thumbnails (front, right, back, left) and returns them under the new `thumbnail_urls` response field. The existing `thumbnail_url` field is unchanged. Adds approximately 3 seconds to task latency.

### `May 22`

- The convert and resize functionality from the [Remesh API](/api/remesh) is now available as standalone [Convert API](/api/convert) and [Resize API](/api/resize) endpoints. The deprecated parameters on Remesh will continue to work for backward compatibility.

### `May 13`

- Added `multi_view_thumbnails` parameter to the [Image to 3D](/api/image-to-3d) API. When set to `true`, the task additionally renders four cardinal-view thumbnails (front, right, back, left) and returns them under the new `thumbnail_urls` response field. The existing `thumbnail_url` field is unchanged. Adds approximately 3 seconds to task latency.

### `May 12`

- Added `resize_longest_side` parameter to the [Remesh](/api/remesh) API. Resizes the model so the longest bounding-box dimension equals the specified value in meters. Mutually exclusive with `resize_height` and `auto_size`.

### `May 11`

- Deprecated `symmetry_mode` parameter on [Text to 3D](/api/text-to-3d), [Image to 3D](/api/image-to-3d), and [Multi-Image to 3D](/api/multi-image-to-3d) endpoints. This parameter no longer affects output.

### `May 7`

- New [Analyze Printability API](/api/analyze-printability): Analyze a succeeded Meshy task — or any 3D model URL — for FDM 3D printing readiness. Async create + stream pattern: POST returns a task_id, then GET / SSE stream return watertightness, volume, hole and non-manifold-edge metrics once SUCCEEDED. Free, no credits consumed.
- New [Repair Printability API](/api/repair-printability): Repair a 3D model for FDM printability — fix non-manifold edges, degenerate faces, holes, and other topology issues. Accepts an existing Meshy `input_task_id` or a `model_url` (`.glb`, `.stl`, `.obj`). Output format matches the input. 10 credits per task.

---

## April 2026
### `Apr 27`

- Added `consumed_credits` field to all task responses across every API endpoint. Shows the number of credits consumed by each task. The field is present for tasks in `PENDING`, `IN_PROGRESS`, and `SUCCEEDED` statuses. Returns `0` for `FAILED` tasks (credits are automatically refunded on failure). See [Pricing](/api/pricing) for credit costs per endpoint.

### `Apr 24`

- Added `model_url` parameter to the [Multi-Color Print](/api/multi-color-print#create-a-multi-color-3d-print-task) endpoint. Submit a textured 3D model directly (via URL or Data URI) instead of an `input_task_id` from a prior Meshy task. Supported formats: `.glb`, `.fbx`. When both `model_url` and `input_task_id` are provided, `input_task_id` takes priority.

### `Apr 21`

- Added `input_task_id` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) endpoints. Reference a completed Text to Image or Image to Image task's output directly instead of providing `image_url` / `image_urls`. When both are supplied, `input_task_id` takes priority.

### `Apr 20`

- Added `hd_texture` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task), and [Retexture](/api/retexture#create-a-retexture-task) endpoints. When enabled, the base color texture is generated at 4K (4096×4096) resolution for higher detail. Only supported for `meshy-6` and `latest`. Defaults to `false`.
- The PBR bundle returned when `enable_pbr` is `true` now includes an `emission` map under `texture_urls` for `meshy-6` and `latest` tasks across [Image to 3D](/api/image-to-3d), [Multi-Image to 3D](/api/multi-image-to-3d), [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task), and [Retexture](/api/retexture). No request change is required.

### `Apr 14`

- Added `decimation_mode` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task), and [Remesh](/api/remesh#create-a-remesh-task) endpoints. Sets the adaptive decimation polycount level: `1` (ultra), `2` (high), `3` (medium), or `4` (low).

### `Apr 12`

- Restructured API documentation: added parameter grouping with dependent fields, inline defaults, required/deprecated badges, expandable sub-navigation, and split Multi-Image, Rigging, and Animation into separate pages.

---

## March 2026
### `Apr 2`

- Added `3mf` format support to `target_formats` parameter across all endpoints ([Image to 3D](/api/image-to-3d), [Text to 3D](/api/text-to-3d), [Remesh](/api/remesh), [Retexture](/api/retexture)). Note: 3MF is opt-in only and must be explicitly requested.
- New [Multi-Color Print API](/api/multi-color-print): Convert 3D models to multi-color 3MF format for 3D printing. Supports 1-16 color palettes with configurable precision. 10 credits per task.

### `Mar 20`

- Retired the Meshy-4 AI model. All API requests using `meshy-4` are no longer supported. Please migrate to `meshy-6` or `latest`.
- Added `auto_size` and `origin_at` parameters to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D](/api/text-to-3d), and [Remesh](/api/remesh#create-a-remesh-task) endpoints. When `auto_size` is enabled, the service uses AI vision to estimate real-world height and resize the model automatically. `origin_at` sets the origin position (`bottom` or `center`). In the Remesh API, `auto_size` is mutually exclusive with `resize_height`.

### `Mar 17`

- Added optional `target_formats` parameter to [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D](/api/text-to-3d), and [Retexture](/api/retexture#create-a-retexture-task) endpoints. Specify which 3D formats to generate (e.g., `["glb", "fbx"]`) to reduce task completion time.

---

## February 2026
### `Feb 28`

- Added `remove_lighting` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task), and [Retexture](/api/retexture#create-a-retexture-task) APIs. Removes highlights and shadows from the base color texture for cleaner results under custom lighting. Only supported for `meshy-6` and `latest`. Defaults to `true`.
- The [Retexture](/api/retexture#create-a-retexture-task) and [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task) APIs now support `meshy-6` as an `ai_model` value for full Meshy 6 texturing.

### `Feb 25`

- Added `image_enhancement` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) APIs for users who want to opt out of input image optimization and preserve the exact appearance of their input. Only supported for `meshy-6` and `latest`. Defaults to `true`.

### `Feb 20`

- Added `model_type` parameter to the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) API for generating low-poly meshes optimized for cleaner polygons.

---

## January 2026
### `Jan 27`

- Added detailed [Failure Modes](/api/errors#task-errors) documentation for all API endpoints, including common HTTP error codes (400, 401, 402, 404, 422, 429) and task failure reasons. This helps developers better understand and handle API errors.

### `Jan 26`

- The [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) API now supports Meshy-6 for full mesh generation. The `ai_model` parameter now accepts `meshy-6` and `latest` resolves to Meshy 6.

### `Jan 22`

- Deprecated `art_style` for the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) API (Meshy-6). The parameter was designed for legacy models (Meshy-4 / Meshy-5) and will be removed in a future release.
- Deprecated `enable_pbr` for the [Text to 3D](/api/text-to-3d) and [Image to 3D](/api/image-to-3d) APIs when `ai_model` is `meshy-4`. This parameter will be removed in a future release.
- Updated the default value of `should_remesh` parameter for Meshy-6 in the [Text to 3D](/api/text-to-3d) and [Image to 3D](/api/image-to-3d) APIs. The parameter now defaults to `false` for `meshy-6`, and `true` for other models if not specified.

### `Jan 19`

- The `latest` option for `ai_model` in the [Text to 3D (Preview)](/api/text-to-3d#create-a-text-to-3d-preview-task) and [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) APIs now resolves to Meshy 6.

### `Jan 12`

- Added a new `model_type` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) API for generating low-poly meshes optimized for cleaner polygons.

---

## December 2025
### `Dec 31`

- Launched the [API Playground](https://www.meshy.ai/api-playground), a dedicated space for developers to explore API parameters and test requests directly on the Meshy website.
- Introduced the [Text to Image API](/api/text-to-image) and [Image to Image API](/api/image-to-image), enabling AI-powered image generation from text prompts and image editing from reference images. Both APIs support the `nano-banana` and `nano-banana-pro` models, with optional multi-view generation.

### `Dec 22`

- The `video_url` field in the [Text to 3D](/api/text-to-3d) API response is now deprecated and will be removed in a future release.

### `Dec 04`

- Added a new `pose_mode` parameter to the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task), [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) APIs. This parameter accepts `a-pose`, `t-pose`, or an empty string (default). The `is_a_t_pose` parameter is now deprecated in favor of `pose_mode`.

---

## November 2025
### `Nov 24`

- The [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) APIs add an optional `save_pre_remeshed_model` parameter and expose `model_urls.pre_remeshed_glb` when a pre-remesh backup is requested.

### `Nov 06`

- The [Text to 3D Refine API](/api/text-to-3d#create-a-text-to-3d-refine-task) now supports `ai_model = latest`, which resolves to Meshy 6 Preview.
- The [Image to 3D API](/api/image-to-3d#create-an-image-to-3d-task) now defaults to Meshy-6-preview texturing when `ai_model = latest` and `should_texture = true`.
- The [Multi-Image to 3D API](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) now supports `ai_model = latest`, running Meshy-6-preview for texturing by default while mesh generation remains Meshy-5.
- The [Retexture API](/api/retexture) adds a `latest` option for `ai_model` (Meshy-6-preview) and now defaults to it when omitted.

### `Nov 04`

- The [Remesh API](/api/remesh#create-a-remesh-task) now preserves textures for uploaded models.
- The [Retexture API](/api/retexture) now preserves textures for uploaded models when the `enable_original_uv` option is enabled.

---

## October 2025
### `Oct 28`

- Added the `x-api-version` response header to indicate the current API server version.

### `Oct 20`

- Added an optional `convert_format_only` boolean parameter to the [Remesh API](/api/remesh#create-a-remesh-task) to support converting the format of the input model file only.
- Added `rigged_character_glb_url` to the response of [The Rigging Task Object](/api/rigging#the-rigging-task-object).

### `Oct 01`

- Remove `text-to-voxel` APIs

---

## September 2025
### `Sep 23`

- Added a `latest` option for `ai_model` in the [Text to 3D](/api/text-to-3d#create-a-text-to-3d-preview-task) API to use Meshy 6 Preview.
- Temporary 50% discount in place for Meshy-6-preview generation tasks to 10 credits that will last until Sep 30, 2025. After the discount period, the cost of Meshy-6-preview tasks will return to the normal 20 credits.

### `Sep 18`

- Updated [API pricing](/api/pricing) for [Text to 3D](/api/text-to-3d) and [Image to 3D](/api/image-to-3d) to reflect different costs for Meshy 6 and other models.

### `Sep 16`

- Added a `latest` option for `ai_model` in the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) API to use Meshy 6 Preview.

### `Sep 4`

- Expanded `model_url` input support to `.glb`, `.gltf`, `.obj`, `.fbx`, `.stl` for [Remesh](/api/remesh#create-a-remesh-task) and [Retexture](/api/retexture).

---

## August 2025
### `Aug 18`

- Added an optional `is_a_t_pose` parameter to the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task), [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) APIs to generate models in an A/T pose.

### `Aug 13`

- Meshy 5 is now stable for [Text to 3D](/api/text-to-3d) and [Image to 3D](/api/image-to-3d) (`ai_model`: `meshy-5`), delivering improved quality and consistency.

---

## July 2025
### `Jul 31`

- Support PBR texture maps in API via `enable_pbr` parameter for latest `meshy-5` model for [Text to 3D](/api/text-to-3d), [Image to 3D](/api/image-to-3d), [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), and [Retexture](/api/retexture) endpoints.

---

## June 2025
### `Jun 25 `
- Added the [Retexture API](/api/retexture), which allows users to retexture 3D models based on Meshy's latest foundation AI models.

### `Jun 19`

- Added deletion APIs for [Text to 3D](/api/text-to-3d), [Image to 3D](/api/image-to-3d), [Remesh](/api/remesh), [Rigging](/api/rigging), and [Animation](/api/animation).
- Added the `ai_model` parameter to the [Text to 3D Refine API](/api/text-to-3d#create-a-text-to-3d-refine-task).

### `Jun 17`

- Added documentation for [Webhooks](/api/webhooks).

### `Jun 10`

- The [Remesh API](/api/remesh#create-a-remesh-task) now supports base64-encoded GLB format models in the `model_url` parameter via Data URI.

---

## May 2025
### `May 20`

- Added an optional `moderation` parameter to the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task), [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task), [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task), and [Multi-Image to 3D](/api/multi-image-to-3d#create-a-multi-image-to-3d-task) APIs. When enabled, input content is automatically screened for potentially harmful content before generation.

### `May 15`

- Introduced the Auto-rigging & Animation API, enabling users to automatically rig and animate 3D models. [Learn more about Rigging](/api/rigging) and [Animation](/api/animation).

---

## April 2025
### `Apr 29`

- Introduced the [Multi-Image to 3D API](/api/multi-image-to-3d#create-a-multi-image-to-3d-task), allowing generation of 3D models from 1 to 4 input images using the `meshy-5` AI model.
- Added the `meshy-5` AI model option to the [Text to 3D](/api/text-to-3d) and [Image to 3D](/api/image-to-3d) APIs. The `latest` tag now also resolves to `meshy-5`.

### `Apr 17`

- Added the `texture_image_url` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task) APIs, allowing users to guide the texture generation process with an image.

---

## March 2025
### `Mar 28`

- Added the `latest` parameter to the [Text to 3D](/api/text-to-3d#create-a-text-to-3d-preview-task) and [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) APIs, enabling access to our upcoming advanced AI models for improved generation quality.

### `Mar 14`

- The [Text to Texture API](/api/text-to-texture#create-a-text-to-texture-task) now supports base64-encoded models in the `model_url` parameter via Data URI, similar to the `image_url` in the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) API.

### `Mar 06`

- Updated pricing for API generation tasks:
  - Text to 3D (Preview): Increased from 2 to 5 credits per call
  - Text to 3D (Refine): Increased from 5 to 10 credits per call
  - Image to 3D: Now 5 credits without texture, 15 credits with texture
  - Text to Texture: Increased from 5 to 10 credits per call
  - Text to Voxel: Remains at 5 credits per call

---

## February 2025
### `Feb 18`

- Added a test mode API key that allows developers to test API integration without consuming credits. All valid requests using this key will return the same sample task results.

### `Feb 13`

- **Breaking Changes**:
  - Free tier task creation will end on `March 20, 2025`. After that, all API task requests will require a paid subscription. To keep your access smooth, we recommend upgrading before the deadline.
  - Use coupon code `APIACCESS` to enjoy a `40%` discount. Thank you for being with us!

- Added Server-Sent Events (SSE) streaming endpoints for real-time task updates:
  - [Text to 3D Stream API](/api/text-to-3d#stream-a-text-to-3d-task)
  - [Image to 3D Stream API](/api/image-to-3d#stream-an-image-to-3d-task)
  - [Remesh Stream API](/api/remesh#stream-a-remesh-task)
  - [Text to Texture Stream API](/api/text-to-texture#stream-a-text-to-texture-task)

### `Feb 6`

- Added the `texture_prompt` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task) APIs, allowing users to guide the texture generation process with a text prompt.

---

## January 2025
### `Jan 23`

- Added endpoints for listing tasks to the [Text to 3D](/api/text-to-3d#list-text-to-3d-tasks), [Image to 3D](/api/image-to-3d#list-image-to-3d-tasks), [Remesh](/api/remesh#list-remesh-tasks) and [Text to Texture](/api/text-to-texture#list-text-to-texture-tasks) APIs.

### `Jan 14`

-	Deprecated the legacy Text to 3D and Image to 3D APIs powered by Meshy-3 AI models.

### `Jan 07`

- Added the `symmetry_mode` parameter to the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) and [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) APIs, allowing for configurable symmetry settings.

---

## December 2024
### `Dec 19`

- Added the [Remesh APIs](/api/remesh), which allow users to remesh and export existing 3D models generated by other Meshy APIs into various formats.
- Updated the polycount limits for our APIs to a range of 100-300,000 for Premium users.

### `Dec 12`

- Deprecated the `model_url` property in all response objects.
- Separated legacy Meshy-3 API from the latest Meshy-4 API.

### `Dec 10`

- Added an `enable_pbr` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Text to 3D Refine](/api/text-to-3d#create-a-text-to-3d-refine-task) APIs.
- Added a `should_texture` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) API.
- Deprecated the `pbr` option in `art_style` parameter for the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) API when using the `meshy-4` AI model.
- Deprecated the `low-poly` option in `art_style` parameter for the [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) API.

### `Dec 5`

- Added a [Get Balance API](/api/balance) to allow users to retrieve their credit balance.
- Added a `should_remesh` parameter to the [Image to 3D](/api/image-to-3d#create-an-image-to-3d-task) and [Text to 3D Preview](/api/text-to-3d#create-a-text-to-3d-preview-task) APIs.

### `Dec 3`

- External APIs now use the `/openapi` prefix. Legacy paths remain supported, but we recommend switching to the new paths.

---

## November 2024
### `Nov 14`

- The [Image to 3D API](/api/image-to-3d#create-an-image-to-3d-task) now supports base64-encoded images.

---