Run a WordPress chatbot on Replicate-hosted models
SleekAI plugs into Replicate’s OpenAI-compatible chat completions endpoint, so a chatbot grounded in WordPress content can run on any Replicate-hosted open model or your own fine-tune. Bring the Replicate API token, SleekAI handles the WordPress wiring.
Bring your own fine-tuned model on Replicate
Replicate’s appeal has always been the long tail. Beyond hosting popular open models like Llama 3 and Mistral, it lets you fine-tune a model on your own data and serve it behind a stable URL with a simple API token. For WordPress sites that want a chatbot tuned to their specific tone, jargon, or product names without standing up an inference cluster from scratch, that pattern is hard to beat.
Replicate exposes an OpenAI-compatible chat completions endpoint at https://api.replicate.com/v1. SleekAI’s OpenAI-compatible provider drops onto it: paste the base URL, your Replicate API token in the key field, and pick the model identifier for the deployment – either a built-in like meta/meta-llama-3.1-70b-instruct or your own fine-tune under your-username/your-model-name.
Because Replicate bills by compute time rather than per token, the economics work well for bursty WordPress traffic. A docs site that does 5,000 chats one week and 50,000 the next pays for exactly the seconds of GPU time used. For sites that want a chatbot in the company’s voice without managing GPUs, the Replicate + SleekAI combination is a sensible default.
Features
Your fine-tune, your bot
Fine-tune a model on Replicate using your product docs, brand voice, and FAQ archive, then point SleekAI at the resulting model string. The WordPress chatbot starts answering in the tone you tuned for, with no extra deployment work.
Wide public model library
Replicate also hosts a long list of public open-weight models. Llama 3.1 70B, Mistral, and others are available under stable namespaces, ready to use through the same OpenAI-compatible chat endpoint and the same API token.
Compute-time billing
Replicate bills by seconds of GPU time used per request. Bursty WordPress traffic – documentation launches, news spikes, seasonal sales – ends up cheaper than per-token pricing on most fine-tune setups, since you only pay for what ran.
Where Replicate plus SleekAI fits
Research and niche models
Academic sites, developer tools, and vertical SaaS deploy specialized models on Replicate – code, biomedicine, legal – and surface them as a chatbot grounded in their published content.
Bursty traffic patterns
Product launches, ticket releases, and viral content moments drive chat volume spikes. Replicate’s compute-time pricing handles those bursts more linearly than fixed-tier per-token pricing.
Repeat questions
Answer the questions visitors ask every day from the pages and posts where you already answer them.
Wire SleekAI to Replicate in four steps
Train or pick a model
Either pick a public model on Replicate like meta/meta-llama-3.1-70b-instruct, or fine-tune your own on training data you provide. Once the fine-tune finishes, note the stable model identifier your-username/your-model-name.
Map your data
Choose the posts, fields, taxonomies or users the bot should know and reference them in the instruction.
Configure the provider in SleekAI
Pick the model identifier per bot – public model or your fine-tune – in each bot configuration screen.
Set display conditions
Decide which pages, post types, roles or logged-in states show the chatbot.
Why Replicate plus WordPress fits brand voice
Off-the-shelf frontier models are very good at the average chat task and only okay at sounding like a specific brand.
A SaaS that prides itself on a punchy, jargon-light tone, a magazine that has spent twenty years polishing its house style, or a developer tools company with strong opinions about how to explain a CLI – none of these get the full benefit of a chatbot that defaults to a generic helpful-assistant register.
Replicate gives those teams a path to a model that actually sounds like them. Drop the fine-tuned model identifier into a bot configuration, pick which post types and meta keys flow into the prompt, set display conditions per template, and the chatbot answering on the site sounds like the rest of the site. The chatbot stops feeling like a bolt-on and starts feeling like part of the publication.
Frequently asked questions
Does SleekAI talk to Replicate’s prediction API directly?
No. Replicate’s classic prediction API is asynchronous and request-specific.
How do I use a fine-tuned model?
Once your fine-tune finishes training on Replicate, it gets a stable model identifier of the form your-username/your-model-name.
Where do I get the Replicate API token?
Sign in at replicate.com, open Account Settings, and create an API token.
Which public models work well for WordPress chat?
meta/meta-llama-3.1-70b-instruct is a solid default for grounded long answers. Smaller Llama and Mistral variants are good for high-volume support chat. For code-heavy bots, the various coder-tuned open models hosted on Replicate work well through the same endpoint.
How does Replicate’s compute-time billing actually work?
For chat use cases, keeping the model warm with steady traffic is the cheapest pattern overall.
Can I run multiple Replicate models on one WordPress install?
Yes. Multibot lets each chatbot pick its own Replicate model string. A common pattern is a fine-tuned brand voice model on the marketing site and a public Llama 3.1 70B on the docs site, both authenticated with the same Replicate API token.
Which AI providers does it support?
OpenAI, Anthropic, Google and OpenRouter, plus custom OpenAI-compatible endpoints. You bring your own API key and the provider bills you directly.
Where does the chatbot get its answers?
From the WordPress data you map into its instruction: post titles and content, post meta, ACF and Meta Box fields, taxonomies, terms, users and user meta, or a WP_Query. Values resolve on every request.
Pricing
Explore our flexible licensing options tailored to your needs. Upgrade your license anytime to access more features, or opt for a lifetime license for ongoing value, including lifetime updates and lifetime support. Our hassle-free upgrade process ensures that our platform can grow with you, starting from whichever plan you choose.
…or get the Bundle Deal
and save €250
The Bundle (unlimited sites)
Pay once, own it forever
Elevate your WordPress site with our exclusive plugin bundle that includes all of our premium plugins in one package. Enjoy lifetime updates and lifetime support. Save significantly compared to buying plugins individually.
What’s included
SleekAI
SleekByte
SleekMotion
SleekPixel
SleekView
SleekRank
SleekTrack