@orka-js/replicate

Replicate adapter for OrkaJS — run any open-source model hosted on Replicate (Llama 2, SDXL, etc.).

Installation

npm install @orka-js/replicate

Quick Start

import { ReplicateAdapter } from '@orka-js/replicate'
import { Orka } from '@orka-js/core'

const llm = new ReplicateAdapter({
  apiToken: process.env.REPLICATE_API_TOKEN!,
  model: 'meta/llama-2-70b-chat',
})

const orka = new Orka({ llm })

const response = await orka.chat([
  { role: 'user', content: 'Hello!' }
])
console.log(response.content)

Streaming

const stream = await llm.stream([
  { role: 'user', content: 'Write a haiku about the ocean' }
])

for await (const chunk of stream) {
  process.stdout.write(chunk.delta ?? '')
}

Configuration

| Option | Type | Default | Description | |--------|------|---------|-------------| | apiToken | string | — | Replicate API token (required) | | model | string | — | Model identifier (required), e.g. 'meta/llama-2-70b-chat' | | temperature | number | 0.7 | Sampling temperature | | maxTokens | number | — | Max tokens in response |

Supported Models

Any model available on replicate.com that supports text generation. Common examples:

| Model | Description | |-------|-------------| | meta/llama-2-70b-chat | Meta Llama 2 70B | | mistralai/mistral-7b-instruct-v0.2 | Mistral 7B | | Any Replicate model slug | Custom hosted models |

API

ReplicateAdapter

const adapter = new ReplicateAdapter(config)

adapter.chat(messages: ChatMessage[])   // Promise<ChatResponse>
adapter.stream(messages: ChatMessage[]) // AsyncIterable<StreamEvent>

Implements: LLMAdapter, StreamingLLMAdapter

Related Packages

• @orka-js/core — Core types and LLMAdapter interface
• @orka-js/ollama — Local models with Ollama
• orkajs — Full bundle