← Image

Leonardo AILeonardo AI

Leonardo（2024-07 被 Canva 收购）Leonardo (acquired by Canva, Jul 2024)

免费+订阅Free + subscription 游戏资产向Game assets Element训练Element training 模型聚合Model aggregation Canva旗下Part of Canva partial lifecycle: active 核验Checked 2026-10-22

Leonardo 是游戏资产向的图像创作台：自研 Phoenix/Lucid 线 + Element 小样本训练做风格一致性 + model hub 聚合选模型；2024-07 起归 Canva 旗下。

Leonardo is the game-asset-oriented image console: self-developed Phoenix/Lucid lines + Element few-shot training for style/character consistency + a model hub; under Canva since July 2024.

适合与不适合When to use it, when not to

适合：游戏概念图/贴图批量、Element 训练锁风格角色、聚合平台内换模型试。不适合：开放权重离线、脱离账号、单模型深度调参。

Fits: game concept art/texture batches, Element training for style/character locks, switching models inside an aggregation. Not a fit: open-weights offline work, deep per-model parameter control.

中文生态China availability 国产 / 国内可直接用 / 备案状态 · 快照domestic / direct access / CAC filing · snapshot 2026-10-22

出品属地Origin: 国外出品foreign-made　Leonardo（悉尼创立，2024-07 起归 Canva，双方均为澳洲主体叙事）。Leonardo (founded in Sydney; under Canva since July 2024 — both Australian-entity narratives).
国内使用Direct access: 直连不可达direct blocked　leonardo.ai 大陆直连实测 000（2026-10-22，代理下 403=CF 拦截可达）——直连不可达，需网络环境。leonardo.ai measured 000 direct from mainland (2026-10-22; 403 via proxy = CF challenge, reachable)— unreachable without a network path.
备案状态CAC filing: 未见于备案名单not in filing list：未见于网信办公示名单（2026-10-22 检索）；该产品未面向大陆公众提供服务，备案不适用。not in the CAC public registry (searched 2026-10-22); the product does not serve the mainland public, so filing does not apply.

备案信息以网信办《生成式人工智能服务已备案信息》公示名单为准（按批次滚动更新，本站只抄编号不解读）；「大陆可直连」是本站从大陆网络直连实测的结果，不是厂商承诺。GitHub 类站点当日实测有波动，口径写在依据里。Filing status is quoted from the CAC public registry (updated batch by batch; this site copies the number and does not interpret it). "Direct access" is this site's own reachability measurement from a mainland network, not a vendor promise. GitHub-hosted sites fluctuated on the measurement day — the note says so.

固定坐标系Fixed coordinate system

8 个维度dimensions, 与同分区其他对象逐项可比each comparable item by item against others in the same section.

模型与开放条件Model & openness: 自研模型线（Phoenix/Lucid 前沿图像模型，官网叙事）+ model hub 聚合第三方模型（官网导航所见）；无开放权重线（未见官方发布）。
In-house model line (Phoenix/Lucid frontier image models, official-site narrative) + a model hub aggregating third-party models (as seen in site navigation); no open-weights line (no official release seen).
运行位置Where it runs: 网页（leonardo.ai）+ API（官网有 API 入口，细节未逐条核验）；生成在厂商云端。
Web (leonardo.ai) + API (an API entry on the official site, details not verified item by item); generation on the vendor's cloud.
本地文件Local files: 作品与训练产物（Element）存账户侧云端；未见官方声明的本机数据面或导出面—— 训练产物能否带走，官方页未见明确说法（未核验）。
Works and training artifacts (Elements) stored account-side in the cloud; no officially stated local data surface or export surface — whether training artifacts can be taken away, no clear statement on official pages (unverified).
关机后的任务Tasks after shutdown: 生成即时/排队（按账号档位波动，未逐条核验）；无「关机续跑」任务面（未核验到声明）；训练任务提交后在云端跑，可离开页面等完成（官网 FAQ 叙事，细节未逐条核验）。
Generation is instant/queued (varies by account tier, not verified item by item); no "keep-running-after-shutdown" task surface (no claim verified); training jobs run on the cloud after submission and you can leave the page and wait for completion (official FAQ narrative, details not verified item by item).
工具与扩展Tools & extensions: Element 训练（官方命名的小样本微调线，风格/角色一致性）、实时画布、model hub 模型市场、游戏资产向工作流（官网叙事：概念图/贴图/资产批量）；Motion 视频线（官网导航所见）。
Element training (the officially named few-shot fine-tuning line, style/character consistency), a real-time canvas, the model hub marketplace, game-asset-oriented workflows (official narrative: concept art/textures/asset batching); the Motion video line (as seen in site navigation).
上下文与记忆Context & memory: 以项目组织；跨会话记忆未核验到官方声明。Element 训练是它的「一致性」机制——训练产物可复用不是记忆。
Organized by projects; cross-session memory — no official statement verified. Element training is its "consistency" mechanism — reusable training artifacts are not memory.
权限与限制Permissions & limits: 账号+订阅体系（D9：价格不采）；训练用途开关未逐条核验。 2024-07 被 Canva 收购（Canva 官方新闻室公告）——收购后数据政策是否随 Canva 统一，未核验。
Account + subscription system (D9: prices not collected); the training-use switch not verified line by line. Acquired by Canva in 2024-07 (Canva official newsroom announcement) — whether data policy unified under Canva post-acquisition, unverified.
适合什么任务Best for: 适合：游戏资产/概念图工作流、Element 训练做风格与角色一致性、聚合平台内选模型。不适合：要开放权重离线跑、要脱离账号、要单模型深度参数控制。
Fits: game-asset/concept-art workflows, Element training for style and character consistency, picking models inside an aggregating platform. Does not fit: open weights running offline, leaving the account, deep per-model parameter control.

头号误解Top misconception

「model hub 聚合第三方模型」≠「能在本地跑这些模型」——聚合的是云端生成服务，不是权重下载'Model hub aggregates third-party models' ≠ 'you can run them locally' — what's aggregated is cloud generation, not weight downloads
归属已变：2024-07 被 Canva 收购——「Leonardo 独立公司」的旧认知过时；收购后条款/数据政策走向未逐条核验Ownership changed: under Canva since July 2024 — 'Leonardo the independent company' is outdated; post-acquisition terms/data-policy alignment not verified line by line
leonardo.ai 对 curl 返回 CF 403——本站核验方法下「读不到」≠「站点失效」leonardo.ai returns CF 403 to curl — 'unreadable by our method' ≠ 'site down'

未知项清单Known unknowns

收购后条款与数据政策是否随 Canva 统一（未核验）
API 的计费与配额面（D9：价格不采；配额行为未逐条核验）
训练用途开关路径（未逐条核验）
Phoenix/Lucid 各档的能力差异（未逐条核验）

Whether terms and data policy are unified under Canva post-acquisition (unverified)
The API's billing and quota surface (D9: prices not collected; quota behavior not verified item by item)
The training-use switch path (not verified line by line)
Capability differences across Phoenix/Lucid tiers (not verified item by item)

证据来源Evidence sources

判断可回到以下一手源复核。本站核验日Every judgement can be rechecked against the primary sources below. Verified on 2026-10-22，内容更新日content updated on 2026-10-22。

类型Type	名称Name	链接Link
official	Leonardo · 官网（产品与能力叙事）Leonardo · official site (product and capability narrative)	https://leonardo.ai
changelog	Leonardo · News（产品更新线，changelog；直抓 CF 403）Leonardo · News (product update line, changelog; direct fetch hit CF 403)	https://leonardo.ai/news
official	Canva Newsroom · 收购公告（2024-07，官方；直抓 CF 403）Canva Newsroom · acquisition announcement (2024-07, official; direct fetch hit CF 403)	https://www.canva.com/newsroom/news/canva-acquires-leonardo/

实测记录Measured runs

本档未跑生成质量与速度测试；结论只覆盖能力面与许可面。

实测建议（三条，都指向官方可读页没给的东西）：

Element 训练的实际样本量与一致性稳定度（对比本地 LoRA 训练）
model hub 里第三方模型的生成质量差异
Canva 收购后账号/作品数据的迁移与政策走向

This site has not completed measured runs. The coding-oriented test protocol is at protocols/coding/_protocol.md (the coding task suite does not apply to creative products).

Suggested measurements (three, all pointing at things the official readable pages do not give):

Element training's actual sample size and consistency stability (compared against local LoRA training)
Generation-quality differences among third-party models in the model hub
Account/work data migration and policy direction after the Canva acquisition