{"id":2189,"date":"2026-08-13T06:11:32","date_gmt":"2026-08-13T06:11:32","guid":{"rendered":"https:\/\/convly.ai\/?p=2189"},"modified":"2026-08-13T06:11:32","modified_gmt":"2026-08-13T06:11:32","slug":"hugging-face-inference-api-guide","status":"publish","type":"post","link":"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/","title":{"rendered":"API di inferenza di Hugging Face: come funziona, quanto costa e quando utilizzarla"},"content":{"rendered":"<div class=\"convly-tldr\">\n<ul>\n<li>The Hugging Face Inference API \u2014 now officially called <strong>Inference Providers<\/strong> \u2014 routes requests to Groq, Together AI, Fireworks, Cerebras, and others through a single HF token at <code>https:\/\/router.huggingface.co\/v1<\/code>.<\/li>\n<li>Free tier: <strong>$0.10\/month<\/strong> in credits for free accounts,<strong>$2.00\/month<\/strong> for PRO users. HF passes through provider rates with no markup.<\/li>\n<li>The legacy <code>hf-inference<\/code> provider (the original serverless API) now focuses on CPU-class inference; GPU models route to third-party providers with warm capacity.<\/li>\n<li>Use dedicated <strong>Inference Endpoints<\/strong> when you need a private or fine-tuned model, guaranteed GPU capacity, or consistent latency \u2014 from $0.50\/hr for an NVIDIA T4 on AWS.<\/li>\n<\/ul>\n<\/div>\n<p>The Hugging Face Inference API gives developers REST access to hundreds of open-weights models \u2014 LLMs, embedding models, image generators, speech, and classifiers \u2014 without provisioning any infrastructure. You authenticate with one HF token, send requests to Hugging Face&#8217;s routing layer, and it dispatches to whichever underlying provider has the model warm and ready. As of 2025, this service is officially called <em>Inference Providers<\/em>, but the token flow, base URL, and core usage pattern remain the same as the original Inference API.<\/p>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_86 counter-flat ez-toc-counter ez-toc-container-direction\">\n<label for=\"ez-toc-cssicon-toggle-item-6a7d8eb0b3b46\" class=\"ez-toc-cssicon-toggle-label\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #000000;color:#000000\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #000000;color:#000000\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/label><input type=\"checkbox\"  id=\"ez-toc-cssicon-toggle-item-6a7d8eb0b3b46\"  aria-label=\"Toggle\" \/><nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#What_Is_the_Hugging_Face_Inference_API_and_What_Changed\" >What Is the Hugging Face Inference API (and What Changed)<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#How_the_Router_Works\" >How the Router Works<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#Authentication_and_Your_First_Request\" >Authentication and Your First Request<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#Free_Tier_Credits_and_What_Happens_Next\" >Free Tier, Credits, and What Happens Next<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#Cold_Starts_and_the_hf-inference_Provider\" >Cold Starts and the hf-inference Provider<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#Inference_API_vs_Inference_Endpoints\" >Inference API vs Inference Endpoints<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#How_Pricing_Compares_with_Other_Providers\" >How Pricing Compares with Other Providers<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#When_Dedicated_Endpoints_or_Self-Hosting_Wins\" >When Dedicated Endpoints or Self-Hosting Wins<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/convly.ai\/it\/hugging-face-inference-api-guide\/#Frequently_Asked_Questions\" >Frequently Asked Questions<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"What_Is_the_Hugging_Face_Inference_API_and_What_Changed\"><\/span>What Is the Hugging Face Inference API (and What Changed)<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Originally, the &#8220;Inference API&#8221; referred to a Hugging Face-hosted serverless service at <code>api-inference.huggingface.co<\/code> that loaded models on demand. That service still exists as the <code>hf-inference<\/code> provider, but as of July 2025 it focuses on CPU-class inference: embeddings, text classification, NER, summarization, and smaller historically significant models like BERT or GPT-2.<\/p>\n<p>For GPU-accelerated inference \u2014 large LLMs, image generation, speech \u2014 Hugging Face now routes through partner providers: Groq, Together AI, Fireworks, Cerebras, DeepInfra, Replicate, Fal AI, and others. The interface is unchanged: one token, one base URL, OpenAI-compatible request format. The router URL is <code>https:\/\/router.huggingface.co\/v1<\/code>. The old <code>api-inference.huggingface.co<\/code> URL still handles legacy calls to <code>hf-inference<\/code>, but new integrations should target the router.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"How_the_Router_Works\"><\/span>How the Router Works<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>When you send a request to <code>router.huggingface.co<\/code>, Hugging Face selects a provider based on a policy you append to the model ID:<\/p>\n<table>\n<thead>\n<tr>\n<th>Policy suffix<\/th>\n<th>Behavior<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>:fastest<\/code> (default)<\/td>\n<td>Highest throughput provider currently available<\/td>\n<\/tr>\n<tr>\n<td><code>:cheapest<\/code><\/td>\n<td>Lowest price per output token<\/td>\n<\/tr>\n<tr>\n<td><code>:preferred<\/code><\/td>\n<td>Your ranked preference list from HF settings<\/td>\n<\/tr>\n<tr>\n<td><code>:groq<\/code>, <code>:together<\/code>, etc.<\/td>\n<td>Force a specific named provider<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Append the policy directly to the model ID string: <code>\"deepseek-ai\/DeepSeek-R1:cheapest\"<\/code>. Omitting a suffix defaults to <code>:fastest<\/code>. Automatic failover is included \u2014 if the selected provider is flagged as unavailable, the router reroutes to an alternative.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Authentication_and_Your_First_Request\"><\/span>Authentication and Your First Request<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Go to <a href=\"https:\/\/huggingface.co\/settings\/tokens\" target=\"_blank\" rel=\"noopener\">huggingface.co\/settings\/tokens<\/a>, create a <strong>Fine-grained<\/strong> token, and enable the <em>Make calls to Inference Providers<\/em> permission. Set it as <code>HF_TOKEN<\/code> in your environment.<\/p>\n<h3>Python \u2014 huggingface_hub<\/h3>\n<pre><code>pip install huggingface_hub<\/code><\/pre>\n<pre><code>import os\nfrom huggingface_hub import InferenceClient\n\nclient = InferenceClient()  # reads HF_TOKEN from environment\n\ncompletion = client.chat.completions.create(\n    model=\"deepseek-ai\/DeepSeek-V3-0324\",\n    messages=[{\"role\": \"user\", \"content\": \"Explain tokenization in two sentences.\"}],\n)\nprint(completion.choices[0].message.content)<\/code><\/pre>\n<h3>Python \u2014 OpenAI drop-in<\/h3>\n<pre><code>from openai import OpenAI\nimport os\n\nclient = OpenAI(\n    base_url=\"https:\/\/router.huggingface.co\/v1\",\n    api_key=os.environ[\"HF_TOKEN\"],\n)\n\ncompletion = client.chat.completions.create(\n    model=\"deepseek-ai\/DeepSeek-V3-0324:fastest\",\n    messages=[{\"role\": \"user\", \"content\": \"Explain tokenization in two sentences.\"}],\n)\nprint(completion.choices[0].message.content)<\/code><\/pre>\n<h3>cURL<\/h3>\n<pre><code>curl https:\/\/router.huggingface.co\/v1\/chat\/completions \n  -H \"Authorization: Bearer $HF_TOKEN\" \n  -H \"Content-Type: application\/json\" \n  -d '{\n    \"model\": \"deepseek-ai\/DeepSeek-V3-0324:fastest\",\n    \"messages\": [{\"role\": \"user\", \"content\": \"Explain tokenization in two sentences.\"}]\n  }'<\/code><\/pre>\n<p>The OpenAI-compatible endpoint covers chat completions only. For other tasks \u2014 text-to-image, embeddings, speech-to-text \u2014 use the <code>huggingface_hub<\/code> Python library or <code>@huggingface\/inference<\/code> JS SDK, which handle provider-specific request formatting automatically.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Free_Tier_Credits_and_What_Happens_Next\"><\/span>Free Tier, Credits, and What Happens Next<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Every Hugging Face account gets a monthly credit allowance that applies automatically to routed requests:<\/p>\n<table>\n<thead>\n<tr>\n<th>Account type<\/th>\n<th>Monthly credits<\/th>\n<th>Pay-as-you-go after?<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Free<\/td>\n<td>$0.10 (subject to change)<\/td>\n<td>Yes \u2014 requires purchasing credits<\/td>\n<\/tr>\n<tr>\n<td>PRO<\/td>\n<td>$2.00<\/td>\n<td>Yes<\/td>\n<\/tr>\n<tr>\n<td>Team \/ Enterprise<\/td>\n<td>$2.00 per seat, pooled<\/td>\n<td>Yes<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Once your credits run out, access does not stop \u2014 you purchase additional credits to continue. Hugging Face charges you the same rate the provider charges, with no additional fee. What a given request costs depends on the model and provider; you can track per-model, per-provider spend at <code>huggingface.co\/settings\/inference-providers\/overview<\/code>.<\/p>\n<p>If you already have accounts with a specific provider, you can set a <em>custom provider key<\/em> in HF settings. Requests still route through HF, but the provider bills you directly and your HF monthly credits do not apply. Before committing to a volume, use the <a href=\"https:\/\/convly.ai\/ai-api-cost-calculator\/\">API cost calculator<\/a> to project monthly spend by model and call volume.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Cold_Starts_and_the_hf-inference_Provider\"><\/span>Cold Starts and the hf-inference Provider<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The legacy <code>hf-inference<\/code> provider loads models on demand. When a model has not been called recently and has gone idle, the first request triggers a model load before the response can begin \u2014 a cold start. This adds noticeable latency to that first call.<\/p>\n<p>As of July 2025, <code>hf-inference<\/code> focuses on CPU inference: embedding models, classifiers, NER, smaller text models. Cold starts are most relevant in this context.<\/p>\n<p>For GPU-accelerated workloads \u2014 large LLMs, image generation \u2014 the router dispatches to third-party providers like Groq or Together AI that run warm, shared GPU capacity. Cold starts are not a concern for these providers in the same way, though you share capacity and have no throughput guarantee during traffic spikes.<\/p>\n<p>If cold-start latency or throughput variability is unacceptable \u2014 say, for a latency-sensitive production endpoint \u2014 a dedicated Inference Endpoint is the solution.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Inference_API_vs_Inference_Endpoints\"><\/span>Inference API vs Inference Endpoints<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The Hugging Face platform has two distinct products for inference. They share the same token system and model Hub, but operate very differently:<\/p>\n<table>\n<thead>\n<tr>\n<th><\/th>\n<th>Inference Providers (API)<\/th>\n<th>Inference Endpoints (Dedicated)<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>Infrastructure<\/strong><\/td>\n<td>Shared, managed by partner providers<\/td>\n<td>Dedicated GPU instance in your chosen region<\/td>\n<\/tr>\n<tr>\n<td><strong>Pricing model<\/strong><\/td>\n<td>Pay-per-request at provider rates<\/td>\n<td>Per-minute billing while running or initializing<\/td>\n<\/tr>\n<tr>\n<td><strong>Cold starts<\/strong><\/td>\n<td>Possible on hf-inference CPU models<\/td>\n<td>None while endpoint is running<\/td>\n<\/tr>\n<tr>\n<td><strong>Private models<\/strong><\/td>\n<td>No \u2014 public Hub models only<\/td>\n<td>Yes \u2014 private repos and fine-tuned models<\/td>\n<\/tr>\n<tr>\n<td><strong>Hardware control<\/strong><\/td>\n<td>None<\/td>\n<td>Choose GPU type, count, and region<\/td>\n<\/tr>\n<tr>\n<td><strong>Minimum cost<\/strong><\/td>\n<td>$0 (within free credits)<\/td>\n<td>~$0.50\/hr running (AWS NVIDIA T4)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Inference Endpoints are billed per minute only while in <em>running<\/em> or <em>initializing<\/em> state \u2014 paused endpoints cost nothing. AWS GPU options range from a T4 at $0.50\/hr (14 GB VRAM) up to an H200 at $5.00\/hr per card (141 GB VRAM). GCP options include H100 at $10.00\/hr per card (80 GB VRAM). Before selecting a tier, use the <a href=\"https:\/\/convly.ai\/llm-vram-calculator\/\">VRAM calculator<\/a> to verify your model fits the target GPU.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"How_Pricing_Compares_with_Other_Providers\"><\/span>How Pricing Compares with Other Providers<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Because Hugging Face routes to the same underlying providers \u2014 Together AI, Fireworks, DeepInfra, Groq \u2014 that you can also reach directly or through OpenRouter, the per-token rates for a given model are generally identical. HF adds no markup.<\/p>\n<p>The practical differences:<\/p>\n<ul>\n<li><strong>Free credits<\/strong>: HF gives $0.10\u2013$2.00\/month in free usage automatically. OpenRouter and direct providers have no equivalent monthly allowance.<\/li>\n<li><strong>Model breadth<\/strong>: HF Inference Providers focuses on open-weights models from the Hub. OpenRouter also covers closed-source models (GPT-4o, Claude, Gemini). If you need both open and proprietary models in one router, OpenRouter covers more ground.<\/li>\n<li><strong>Non-chat tasks<\/strong>: Embeddings, image generation, and speech are available through HF&#8217;s SDK. Most competing routers are chat-completion only.<\/li>\n<li><strong>Billing consolidation<\/strong>: One HF account covers all providers, with a single usage dashboard. Direct provider accounts require separate billing relationships for each.<\/li>\n<\/ul>\n<p>For a full model comparison by price and capability, see the <a href=\"https:\/\/convly.ai\/models\/\">AI models database<\/a> covering specs and pricing across major models.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"When_Dedicated_Endpoints_or_Self-Hosting_Wins\"><\/span>When Dedicated Endpoints or Self-Hosting Wins<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Use Inference Providers when you are prototyping, have variable or unpredictable load, and need access to a broad public model catalog without managing servers.<\/p>\n<p>Switch to a dedicated Inference Endpoint when:<\/p>\n<ul>\n<li>You need to serve a fine-tuned or private model not available on the public Hub.<\/li>\n<li>Latency consistency is a requirement \u2014 shared providers can have variable response times under load.<\/li>\n<li>You need guaranteed throughput for a production SLA.<\/li>\n<\/ul>\n<p>Consider self-hosting when your call volume is high enough that per-token costs exceed the amortized cost of owning hardware, or when data-privacy requirements prohibit sending inputs to third-party APIs. The<a href=\"https:\/\/convly.ai\/self-hosting-vs-api-calculator\/\">self-hosting vs API break-even calculator<\/a> gives a concrete cost comparison based on your request volume and model size. For hardware recommendations if you go that route, see the <a href=\"https:\/\/convly.ai\/best-gpus-for-local-llms-2026\/\">best GPUs for running LLMs locally<\/a>.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Frequently_Asked_Questions\"><\/span>Frequently Asked Questions<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3>What is the difference between the Inference API and Inference Endpoints?<\/h3>\n<p>The Inference API (now Inference Providers) is a shared, pay-per-request service that routes to partner GPU providers and HF&#8217;s own CPU infrastructure. Inference Endpoints are dedicated GPU instances you provision in a cloud region; they run a single model continuously and are billed per minute of uptime. Use the API for prototyping and variable workloads; use Endpoints for production latency guarantees and private or fine-tuned models.<\/p>\n<h3>Do I need a Pro subscription to use the Inference API?<\/h3>\n<p>No. A free account gets $0.10\/month in credits, which is sufficient for light experimentation. PRO subscribers receive $2.00\/month. Both tiers support pay-as-you-go credit purchases once the monthly allowance is exhausted. The PRO plan&#8217;s main inference advantage is the higher monthly credit amount, not gated access to models or providers.<\/p>\n<h3>Why is my first request much slower than subsequent ones?<\/h3>\n<p>If you are routing to the <code>hf-inference<\/code> provider, it loads models on demand. A model that has gone idle must be loaded into memory before your request can complete, adding latency to that first call. This does not apply to GPU providers like Groq or Together AI, which run warm shared infrastructure. Specifying <code>:fastest<\/code> or a named GPU provider in the model ID will avoid this delay entirely.<\/p>\n<h3>Is the HF Inference API compatible with the OpenAI Python SDK?<\/h3>\n<p>Yes, for chat completions. Set <code>base_url=\"https:\/\/router.huggingface.co\/v1\"<\/code> and pass your HF token as <code>api_key<\/code>. The <code>\/v1\/chat\/completions<\/code> and <code>\/v1\/models<\/code> endpoints are OpenAI-compatible. For other task types \u2014 embeddings, image generation, speech \u2014 you need the <code>huggingface_hub<\/code> Python library or <code>@huggingface\/inference<\/code> JS SDK; those are not covered by the OpenAI-compatible endpoint.<\/p>\n<h3>Can I serve a fine-tuned model through the Inference API?<\/h3>\n<p>Not through Inference Providers \u2014 it only serves models available in the public Hub catalog and supported by a partner provider. For a private or fine-tuned model, deploy a dedicated Inference Endpoint, which supports private Hub repos and lets you bring your own model weights to a GPU instance you control.<\/p>\n<h3>How do I track and control costs?<\/h3>\n<p>Your usage breakdown by model and provider is at <code>huggingface.co\/settings\/inference-providers\/overview<\/code>. Team and Enterprise admins can set spending limits and disable specific providers from the organization settings page. For forward-looking cost estimates before you start building, use the <a href=\"https:\/\/convly.ai\/ai-api-cost-calculator\/\">API cost calculator<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>The Hugging Face Inference API \u2014 now officially called Inference Providers \u2014 routes requests to Groq, Together AI, Fireworks, Cerebras, [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":2190,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[7],"tags":[],"class_list":["post-2189","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news"],"_links":{"self":[{"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/posts\/2189","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/comments?post=2189"}],"version-history":[{"count":1,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/posts\/2189\/revisions"}],"predecessor-version":[{"id":2191,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/posts\/2189\/revisions\/2191"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/media\/2190"}],"wp:attachment":[{"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/media?parent=2189"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/categories?post=2189"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/convly.ai\/it\/wp-json\/wp\/v2\/tags?post=2189"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}