{"id":2157,"date":"2026-08-10T14:04:16","date_gmt":"2026-08-10T14:04:16","guid":{"rendered":"https:\/\/convly.ai\/?p=2157"},"modified":"2026-08-10T14:04:16","modified_gmt":"2026-08-10T14:04:16","slug":"koboldcpp-guide","status":"publish","type":"post","link":"https:\/\/convly.ai\/pt\/koboldcpp-guide\/","title":{"rendered":"KoboldCpp: guia completo para o runtime local de LLM em \u00fanico bin\u00e1rio"},"content":{"rendered":"<div class=\"convly-tldr\">\n<ul>\n<li>KoboldCpp is a single executable \u2014 download it, point it at a GGUF model file, and a browser UI plus OpenAI-compatible API start immediately on port 5001.<\/li>\n<li>GPU offload is controlled by <code>--gpulayers N<\/code>; start with 999 to try full offload and reduce if you hit out-of-memory errors.<\/li>\n<li>Use it when you want a built-in story\/chat UI or need KoboldAI-compatible endpoints; use Ollama if you prefer a managed model library and CLI-first workflow.<\/li>\n<li>No install step, no package manager, no daemon \u2014 just a single binary and a GGUF file.<\/li>\n<\/ul>\n<\/div>\n<p>KoboldCpp is a single-file local LLM runtime built on top of llama.cpp. Download one binary, point it at a GGUF model, and you immediately get a browser-based chat UI and an OpenAI-compatible REST API \u2014 no package manager, no daemon to configure, no install step required. It runs on Windows, macOS, and Linux with optional GPU acceleration via CUDA, Metal, Vulkan, or OpenCL.<\/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-6a7a0e021569d\" 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-6a7a0e021569d\"  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\/pt\/koboldcpp-guide\/#Downloading_KoboldCpp\" >Downloading KoboldCpp<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#Getting_a_GGUF_Model\" >Getting a GGUF Model<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#Launching_KoboldCpp\" >Launching KoboldCpp<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#The_Web_UI_and_OpenAI-Compatible_API\" >The Web UI and OpenAI-Compatible API<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#GPU_Offload_Picking_the_Right_%E2%80%93gpulayers_Value\" >GPU Offload: Picking the Right &#8211;gpulayers Value<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#Context_Size_and_Performance_Settings\" >Context Size and Performance Settings<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#KoboldCpp_vs_Ollama_vs_llamacpp\" >KoboldCpp vs Ollama vs llama.cpp<\/a><\/li><li class='ez-toc-page-1'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/convly.ai\/pt\/koboldcpp-guide\/#Frequently_Asked_Questions\" >Frequently Asked Questions<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"Downloading_KoboldCpp\"><\/span>Downloading KoboldCpp<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Releases are published on the <a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\/releases\" rel=\"noopener\" target=\"_blank\">KoboldCpp GitHub releases page<\/a>. Each release ships platform-specific binaries; pick the one that matches your hardware.<\/p>\n<h3>Windows<\/h3>\n<p>Download <strong>koboldcpp.exe<\/strong> for NVIDIA GPU support (CUDA libraries are bundled \u2014 no separate CUDA toolkit install required, only the standard NVIDIA display driver). If you have no NVIDIA GPU, download <strong>koboldcpp_nocuda.exe<\/strong> instead. Double-clicking the .exe opens a graphical launcher where you can browse for your model file and configure settings before starting the server. To skip the launcher and start from the command line, pass the <code>--skiplauncher<\/code> flag.<\/p>\n<h3>macOS<\/h3>\n<p>Download the macOS binary from the releases page (typically named <strong>koboldcpp_mac<\/strong> or distributed as a .dmg). Metal GPU acceleration is included automatically \u2014 no extra flag is needed; KoboldCpp detects Apple Silicon and uses Metal by default. On first launch, macOS may warn that the binary is from an unidentified developer; right-click \u2192 Open to bypass Gatekeeper.<\/p>\n<h3>Linux<\/h3>\n<p>Download the Linux binary and make it executable:<\/p>\n<pre><code>chmod +x koboldcpp\n.\/koboldcpp --model \/path\/to\/model.gguf<\/code><\/pre>\n<p>Prebuilt Linux binaries include CPU and Vulkan support. For CUDA on NVIDIA cards, look for a release asset with a <code>cu<\/code> suffix in the filename, or compile from source with <code>make LLAMA_CUDA=1<\/code>. If your driver is too old for the bundled CUDA version, the Vulkan build is a reliable fallback.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Getting_a_GGUF_Model\"><\/span>Getting a GGUF Model<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>KoboldCpp loads GGUF files directly \u2014 the same format used by llama.cpp and Ollama. The primary source is Hugging Face; search for a model name plus &#8220;GGUF&#8221;. Before downloading, use the <a href=\"https:\/\/convly.ai\/llm-vram-calculator\/\">VRAM calculator<\/a> to confirm the model will fit your GPU at your chosen context size. Quantization tiers to know:<\/p>\n<table>\n<thead>\n<tr>\n<th>Quantization<\/th>\n<th>Quality<\/th>\n<th>Size vs FP16<\/th>\n<th>When to use<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Q2_K<\/td>\n<td>Noticeable loss<\/td>\n<td>~25%<\/td>\n<td>Very limited VRAM only<\/td>\n<\/tr>\n<tr>\n<td>Q4_K_M<\/td>\n<td>Good<\/td>\n<td>~45%<\/td>\n<td>Default choice for most hardware<\/td>\n<\/tr>\n<tr>\n<td>Q5_K_M<\/td>\n<td>Very good<\/td>\n<td>~55%<\/td>\n<td>When you have spare VRAM<\/td>\n<\/tr>\n<tr>\n<td>Q8_0<\/td>\n<td>Near-lossless<\/td>\n<td>~80%<\/td>\n<td>High-VRAM cards or large CPU RAM<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2><span class=\"ez-toc-section\" id=\"Launching_KoboldCpp\"><\/span>Launching KoboldCpp<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The minimal command on any platform:<\/p>\n<pre><code>.\/koboldcpp --model \/path\/to\/model.gguf<\/code><\/pre>\n<p>This starts the server on <strong>http:\/\/localhost:5001<\/strong>. Open that URL in your browser to reach the web UI.<\/p>\n<h3>Windows \u2014 GUI Launcher<\/h3>\n<p>Double-click <code>koboldcpp.exe<\/code>. The launcher window lets you browse for a model file, set GPU layers, context size, and backend without touching the command line. Click <strong>Launch<\/strong> when done; a terminal window opens showing the server log and the browser UI launches automatically.<\/p>\n<h3>Command-Line (all platforms)<\/h3>\n<p>A typical launch command with GPU offload, custom context, and explicit port:<\/p>\n<pre><code>.\/koboldcpp \n  --model .\/models\/llama3-8b-q4_k_m.gguf \n  --gpulayers 32 \n  --contextsize 8192 \n  --port 5001<\/code><\/pre>\n<p>Key flags reference:<\/p>\n<table>\n<thead>\n<tr>\n<th>Flag<\/th>\n<th>Default<\/th>\n<th>What it controls<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>--model &lt;path&gt;<\/code><\/td>\n<td>\u2014<\/td>\n<td>Path to GGUF file (required)<\/td>\n<\/tr>\n<tr>\n<td><code>--gpulayers &lt;n&gt;<\/code><\/td>\n<td>0<\/td>\n<td>Transformer layers offloaded to GPU<\/td>\n<\/tr>\n<tr>\n<td><code>--contextsize &lt;n&gt;<\/code><\/td>\n<td>4096<\/td>\n<td>Maximum context window in tokens<\/td>\n<\/tr>\n<tr>\n<td><code>--port &lt;n&gt;<\/code><\/td>\n<td>5001<\/td>\n<td>HTTP port<\/td>\n<\/tr>\n<tr>\n<td><code>--host &lt;addr&gt;<\/code><\/td>\n<td>127.0.0.1<\/td>\n<td>Bind address (use 0.0.0.0 to expose on LAN)<\/td>\n<\/tr>\n<tr>\n<td><code>--threads &lt;n&gt;<\/code><\/td>\n<td>auto<\/td>\n<td>CPU threads for inference<\/td>\n<\/tr>\n<tr>\n<td><code>--flashattention<\/code><\/td>\n<td>off<\/td>\n<td>Reduces VRAM for long contexts via Flash Attention<\/td>\n<\/tr>\n<tr>\n<td><code>--usecublas<\/code><\/td>\n<td>off<\/td>\n<td>Force CUDA backend (NVIDIA)<\/td>\n<\/tr>\n<tr>\n<td><code>--usevulkan<\/code><\/td>\n<td>off<\/td>\n<td>Vulkan backend (AMD\/Intel\/NVIDIA)<\/td>\n<\/tr>\n<tr>\n<td><code>--skiplauncher<\/code><\/td>\n<td>off<\/td>\n<td>Windows only: bypass the GUI launcher<\/td>\n<\/tr>\n<tr>\n<td><code>--smartcontext<\/code><\/td>\n<td>off<\/td>\n<td>Shift context instead of stopping when full<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2><span class=\"ez-toc-section\" id=\"The_Web_UI_and_OpenAI-Compatible_API\"><\/span>The Web UI and OpenAI-Compatible API<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Once running, KoboldCpp exposes two interfaces from the same port:<\/p>\n<ul>\n<li><strong>Browser UI \u2014 http:\/\/localhost:5001<\/strong>: A full-featured text generation interface with story, chat, and instruct modes. Supports prompt templates, memory, author&#8217;s notes, and world info fields inherited from the KoboldAI project.<\/li>\n<li><strong>KoboldAI API \u2014 http:\/\/localhost:5001\/api\/v1<\/strong>: Used by frontends like SillyTavern and Agnaistic.<\/li>\n<li><strong>OpenAI-compatible API \u2014 http:\/\/localhost:5001\/v1<\/strong>: Implements <code>\/v1\/chat\/completions<\/code> and <code>\/v1\/completions<\/code>. Any client that accepts a custom base URL works, including LangChain, the OpenAI Python SDK, and most open-source chat apps.<\/li>\n<\/ul>\n<p>To point the OpenAI Python SDK at KoboldCpp:<\/p>\n<pre><code>from openai import OpenAI\n\nclient = OpenAI(\n    base_url=\"http:\/\/localhost:5001\/v1\",\n    api_key=\"unused\"  # KoboldCpp does not require an API key by default\n)\nresponse = client.chat.completions.create(\n    model=\"koboldcpp\",\n    messages=[{\"role\": \"user\", \"content\": \"Hello\"}]\n)<\/code><\/pre>\n<h2><span class=\"ez-toc-section\" id=\"GPU_Offload_Picking_the_Right_%E2%80%93gpulayers_Value\"><\/span>GPU Offload: Picking the Right &#8211;gpulayers Value<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Each transformer layer offloaded to GPU moves processing off the CPU and dramatically increases tokens-per-second. The tradeoff is VRAM. A model&#8217;s total layer count is fixed by its architecture \u2014 common values are 32 for 7B\/8B models, 40 for 13B models, and 80 for 70B models. Each layer consumes a roughly equal share of the model&#8217;s total GPU memory.<\/p>\n<p><strong>Practical approach:<\/strong><\/p>\n<ol>\n<li>Use the <a href=\"https:\/\/convly.ai\/llm-vram-calculator\/\">VRAM calculator<\/a> to estimate how many layers fit at your target context size.<\/li>\n<li>Start with <code>--gpulayers 999<\/code> to attempt full offload. KoboldCpp clamps this to the model&#8217;s actual layer count automatically.<\/li>\n<li>If you get an out-of-memory error on startup, reduce the value and retry. The server log prints per-layer allocation to help you calibrate.<\/li>\n<\/ol>\n<p>Partial offload is supported and useful \u2014 even offloading half the layers of a large model gives a significant speed improvement over pure CPU inference. If you are unsure which GPU to pair with a model, see the <a href=\"https:\/\/convly.ai\/best-gpus-for-local-llms-2026\/\">best GPUs for local LLMs guide<\/a> and the <a href=\"https:\/\/convly.ai\/vram-requirements-every-major-llm-2026\/\">VRAM requirements breakdown by model<\/a>.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Context_Size_and_Performance_Settings\"><\/span>Context Size and Performance Settings<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Context size (<code>--contextsize<\/code>) is the single biggest driver of VRAM beyond model weights. A 7B model at Q4_K_M uses roughly 4 GB for weights; extending context from 4096 to 32768 tokens can add several more gigabytes to the KV cache. Enable <code>--flashattention<\/code> to reduce KV cache footprint \u2014 this is particularly effective at very long contexts and costs nothing in output quality.<\/p>\n<p>Other settings that affect speed:<\/p>\n<ul>\n<li><code>--threads<\/code>: for CPU-only inference, set this close to your physical core count, not your logical (hyperthreaded) count.<\/li>\n<li><code>--batchsize<\/code>: larger values (e.g., 512) improve prompt-processing speed at the cost of peak VRAM during prefill.<\/li>\n<li><code>--smartcontext<\/code>: when the context fills up, KoboldCpp shifts the oldest tokens out instead of stopping generation \u2014 useful for long interactive sessions.<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"KoboldCpp_vs_Ollama_vs_llamacpp\"><\/span>KoboldCpp vs Ollama vs llama.cpp<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>All three are built on the same llama.cpp engine and support GGUF models. The differences are in workflow and interface.<\/p>\n<table>\n<thead>\n<tr>\n<th><\/th>\n<th>KoboldCpp<\/th>\n<th>Ollama<\/th>\n<th>llama.cpp (llama-server)<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Distribution<\/td>\n<td>Single binary, no install<\/td>\n<td>Installer + background daemon<\/td>\n<td>Build from source or prebuilt<\/td>\n<\/tr>\n<tr>\n<td>Web UI<\/td>\n<td>Yes, built-in (rich)<\/td>\n<td>None (third-party required)<\/td>\n<td>Minimal<\/td>\n<\/tr>\n<tr>\n<td>Model management<\/td>\n<td>Manual \u2014 bring your own GGUF<\/td>\n<td>Built-in: <code>ollama pull<\/code><\/td>\n<td>Manual \u2014 bring your own GGUF<\/td>\n<\/tr>\n<tr>\n<td>OpenAI-compatible API<\/td>\n<td>Yes (\/v1)<\/td>\n<td>Yes<\/td>\n<td>Yes<\/td>\n<\/tr>\n<tr>\n<td>KoboldAI API<\/td>\n<td>Yes<\/td>\n<td>No<\/td>\n<td>No<\/td>\n<\/tr>\n<tr>\n<td>Best for<\/td>\n<td>Creative writing, roleplay, SillyTavern<\/td>\n<td>Dev tooling, CLI, systemd service<\/td>\n<td>Minimal footprint, custom builds<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><strong>Choose KoboldCpp<\/strong> if you want zero-install setup, the built-in story\/chat UI, or compatibility with KoboldAI frontends like SillyTavern.<br \/><strong>Choose Ollama<\/strong> if you want a managed model library, a <code>systemd<\/code> service, or tighter CLI integration \u2014 see the <a href=\"https:\/\/convly.ai\/what-is-ollama-complete-guide-2026\/\">Ollama complete guide<\/a> for a full walkthrough.<br \/><strong>Choose llama.cpp directly<\/strong> if you are building a custom integration or need the absolute latest upstream features before they reach downstream wrappers.<\/p>\n<p>If you are still deciding whether to self-host at all versus calling a hosted API, the <a href=\"https:\/\/convly.ai\/self-hosting-vs-api-calculator\/\">self-hosting vs API break-even calculator<\/a> can help you model the cost crossover point.<\/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>Does KoboldCpp require installing CUDA drivers separately?<\/h3>\n<p>On Windows, koboldcpp.exe bundles the CUDA runtime libraries, so you only need the standard NVIDIA display driver \u2014 no separate CUDA toolkit installation. On Linux, CUDA builds typically link against the installed CUDA runtime, so driver version compatibility matters; if your driver is too old, the Vulkan build is the easiest fallback.<\/p>\n<h3>What does &#8211;gpulayers 0 mean?<\/h3>\n<p>Zero GPU layers means all computation runs on CPU. This is the default when no GPU flag is set. CPU inference is much slower \u2014 typically 2\u201310 tokens\/second on a modern CPU versus 40\u2013100+ tokens\/second on a mid-range GPU \u2014 but works on any machine regardless of GPU availability.<\/p>\n<h3>Can I use KoboldCpp as an OpenAI API drop-in for my application?<\/h3>\n<p>Yes. Set your OpenAI client&#8217;s <code>base_url<\/code> to <code>http:\/\/localhost:5001\/v1<\/code> and any non-empty string as the <code>api_key<\/code> (it is not validated by default). The <code>model<\/code> field is accepted but ignored \u2014 whichever GGUF is loaded is always used. Chat completions and text completions both work; embeddings and image endpoints are not supported.<\/p>\n<h3>How do I run two different models at the same time?<\/h3>\n<p>Each KoboldCpp process handles one model. Launch a second instance with a different <code>--port<\/code> value (e.g., 5002) pointing at a different GGUF file. There is no built-in load balancer; route between instances at the application layer.<\/p>\n<h3>Why is generation slower than expected even with a GPU?<\/h3>\n<p>The most common cause is partial CPU offload: if <code>--gpulayers<\/code> is lower than the model&#8217;s total layer count, the remaining layers run on CPU and create a bottleneck. Check the startup log \u2014 KoboldCpp prints exactly how many layers went to GPU versus CPU. Also confirm that the correct backend (CUDA\/Metal\/Vulkan) appears in the startup output rather than a CPU fallback.<\/p>\n<h3>Is it safe to expose KoboldCpp on a network?<\/h3>\n<p>By default KoboldCpp binds to <code>127.0.0.1<\/code> (localhost only). To expose it on a LAN, add <code>--host 0.0.0.0<\/code>. There is no built-in authentication, so exposing it to untrusted networks or the public internet is not recommended without a reverse proxy with authentication in front of it.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>KoboldCpp is a single executable \u2014 download it, point it at a GGUF model file, and a browser UI plus [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":2158,"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-2157","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news"],"_links":{"self":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/2157","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/comments?post=2157"}],"version-history":[{"count":1,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/2157\/revisions"}],"predecessor-version":[{"id":2159,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/posts\/2157\/revisions\/2159"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/media\/2158"}],"wp:attachment":[{"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/media?parent=2157"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/categories?post=2157"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/convly.ai\/pt\/wp-json\/wp\/v2\/tags?post=2157"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}