Leonardo AI Showcase Data on Demand
Leonardo AI's public showcase pages are a useful record of how images were prompted, which styles were applied and which models produced them.
Reading those pages by hand does not scale, and a plain HTTP request often returns a shell that only fills in after JavaScript runs. A managed scraping API handles the browser, the proxy rotation and the anti-bot layer, then hands back either rendered HTML or structured JSON that your pipeline can consume.
What a public Leonardo AI showcase page yields
A public showcase page is the unit of collection. Across those pages the API can return the image URL, the prompt text, style tags, generation parameters, the model name and the creator handle. That set is enough for three common jobs: comparing visual strategies, studying which prompt structures and modifiers tend to accompany strong outputs, and watching for generations that reference your own brand or signature style.
Before you build, decide which of those jobs you actually have. Prompt research needs the prompt and parameter fields; brand monitoring needs the creator handle and a way to match names or visual cues; competitive analysis needs the image URL and style tags. Collecting fields you will never query adds cost without adding insight.
Matching request controls to the data you need
The API exposes a small set of request parameters, and each one answers a specific problem. Use this map to choose the minimum configuration for your job rather than enabling everything by default.
- render_js=true — runs the page in a real headless browser and returns the post-render HTML. Needed whenever the showcase content is injected by JavaScript.
- stealth_proxy — routes the request through stealthier infrastructure when a target responds aggressively to normal traffic.
- country_code — routes through a residential IP in the named country, useful when content differs by geography.
- extract_rules — CSS selectors mapped to field names, returning parsed JSON.
- ai_extract_rules — a natural-language specification of the fields you want, also returning parsed JSON.
A practical default for showcase pages is render_js=true plus one extraction mode. Add stealth_proxy only if you see blocks, and country_code only if you have evidence that the content varies by region.
Worked example: one showcase URL to one JSON record
Take a single public showcase URL. Send it with render_js=true so the browser executes the page, and attach extract_rules that map CSS selectors to the field names you want — for example an image selector, a prompt selector, a style-tag selector and a creator-handle selector. The response arrives as parsed JSON, one record per matched element set.
If the page markup is unstable or you do not yet know the selectors, swap extract_rules for ai_extract_rules and describe the fields in plain language. The trade-off is precision: selector rules are deterministic and cheap to re-run, while natural-language rules are faster to author but should be spot-checked against a sample of pages before you trust them at volume. Either way, the output shape is the same JSON contract, so downstream code does not change when you switch modes.
Two different failures, two different fixes
When a Leonardo AI scrape returns nothing useful, the cause is usually one of two things, and they need different remedies.
Blocked traffic. If the response is a challenge page, a CAPTCHA or an outright refusal, the request was identified as automated. Enable stealth_proxy on that request. The API already rotates residential proxies and uses a real headless browser, so ordinary traffic looks normal; stealth_proxy is the escalation for targets that are more aggressive than ordinary.
Empty or partial content. If the response is a valid page but the showcase fields are missing, the JavaScript had not finished when the HTML was captured. Set render_js=true and confirm the selectors target elements that exist after rendering, not placeholders in the initial markup.
Separating these two cases saves time: adding stealth_proxy to a rendering problem does nothing, and adding render_js to a block does not help either.
Throughput, concurrency and pacing
The API applies concurrency limits per plan rather than per-target rate limits. That means your ceiling is the number of simultaneous requests your plan allows, not a fixed number of calls to Leonardo AI per minute. If the target itself throttles aggressive patterns, space requests out or spread them across multiple sessions.
Plan tiers differ in API credits and concurrent requests. The pricing table below preserves the captured plan figures exactly; treat them as a dated reference and confirm current terms before committing to a tier. Credit consumption also depends on which features you enable — JavaScript rendering and premium proxies each carry their own credit cost, so a request that uses both is more expensive than a plain fetch.
Getting from zero to a first request
The setup path is short. Create an account to receive an API key, then either install one of the supplied SDKs (Python, Node, Go, PHP, Ruby) or call the API directly with cURL. Each scrape becomes a single function call or a single HTTP request; there is no proxy pool or headless Chrome instance for you to operate.
If you prefer not to write code at all, the dashboard request builder lets you paste a URL, set parameters and copy a generated snippet in any of the supported languages. That is also the quickest way to validate selectors before wiring them into a production job.
Scope and compliance
This workflow targets public showcase and generation pages. It does not reach private accounts, private generations or anything behind a login. Keep your collection limited to public pages, respect the target's terms and applicable law, and avoid collecting personal data unless you have a lawful basis. The API itself does not store personal data from scraped sites unless the user explicitly requests it.
| Plan | Price / month | API credits | Concurrent requests | Priority support | Account manager | Team management |
|---|---|---|---|---|---|---|
| Hobby | $19 | 75,000 | 25 | Not included | Not included | Not included |
| Freelance | $49 | 250,000 | 50 | Not included | Not included | Not included |
| Startup (Recommended) | $99 | 1,000,000 | 100 | Not included | Not included | Not included |
| Business | $249 | 3,000,000 | 200 | Included | Included | Included |
| Business + | $599 | 8,000,000 | 400 | Included | Included | Included |
| Enterprise 14 | $999 | 14,000,000 | 500 | Included | Included | Included |
| Enterprise 24 | $1,599 | 24,000,000 | 600 | Included | Included | Included |
| Enterprise 41 (Recommended) | $2,399 | 41,000,000 | 700 | Included | Included | Included |
| Enterprise 69 | $3,599 | 69,000,000 | 800 | Included | Included | Included |
| Enterprise 120 | $5,799 | 120,000,000 | 900 | Included | Included | Included |
Captured plan figures, exclusive of VAT. JavaScript rendering, rotating and premium proxies, geotargeting and screenshots/extraction rules are listed as included on every tier. Prices and allowances change; confirm current terms before purchase. API credits are valid for one month and do not roll over; requests start at 1 credit each, JS rendering adds 5 credits per successful request, and premium proxies add 10 or 25 credits depending on whether JS rendering is enabled.
| Goal | Fields to collect | Request control | Notes |
|---|---|---|---|
| Prompt research | prompt text, generation parameters, style tags | extract_rules or ai_extract_rules | Selector rules are deterministic; natural-language rules need spot checks |
| Competitive visual analysis | image URL, style tags, model name | extract_rules | Keep image URL as a stable key |
| Brand mimicry monitoring | creator handle, prompt text, image URL | extract_rules + stealth_proxy if blocked | Match names and visual cues; public pages only |
| Geo-variant content | any of the above | country_code | Only when content is known to differ by country |
| JavaScript-rendered showcase | any of the above | render_js=true | Required for post-render HTML |
The field map is a planning aid derived from the documented request parameters; it does not assert that every field is present on every showcase page.
Frequently asked questions
Will I get blocked when scraping Leonardo AI?
Requests are routed through rotating residential proxies and a real headless browser, so ordinary traffic looks normal. If a target responds aggressively, enable stealth_proxy on that request to escalate.
Can I scrape Leonardo AI without writing code?
Yes. The dashboard request builder lets you paste a URL, set parameters and copy a generated snippet in Python, Node, Go, PHP, Ruby or cURL.
How do I handle JavaScript when scraping Leonardo AI?
Pass render_js=true. The page runs in a real headless browser, the JavaScript executes server-side, and the post-render HTML is returned.
Can I scrape Leonardo AI from a specific country?
Yes. Pass country_code on the request to route through a residential IP in that country. This is useful when content varies by geography.
What's the rate limit for scraping Leonardo AI?
Limits are applied as concurrency per plan rather than per-target rate limits. If the target throttles aggressive patterns, space requests out or use multiple sessions.
How do I get structured JSON from Leonardo AI?
Pass extract_rules with CSS selectors mapped to field names, or use ai_extract_rules with a natural-language specification. The response comes back as parsed JSON.
Should I use extract_rules or ai_extract_rules?
Use extract_rules when the page markup is stable and you know the selectors — it is deterministic and cheap to re-run. Use ai_extract_rules when the markup is unstable or you are still discovering which fields matter, then spot-check the output against a sample before relying on it at volume.