Production AI inference, one endpoint away.
Kymo gives you a fast, OpenAI-compatible chat completions API plus a full playground for prompting, streaming, vision and structured output.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.KYMO_API_KEY,
baseURL: "https://kymo.app/api/public/v1",
});
const res = await client.chat.completions.create({
model: "llama-3.3-70b-versatile",
messages: [{ role: "user", content: "Explain quantum tunneling briefly." }],
stream: true,
});
for await (const chunk of res) {
process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}Everything you need to ship
OpenAI-compatible
Point any OpenAI or Groq SDK at the Kymo base URL. Same request and response shapes, streaming included.
Curated model catalog
Llama 3.3, GPT-OSS 120B, Qwen vision and agentic Compound models, all behind one endpoint.
Vision & multimodal
Send images alongside text for OCR, captioning and visual Q&A directly from the playground or the API.
Structured JSON mode
Force strictly valid JSON responses for extraction pipelines and tool-driven workflows.
Keys with guardrails
Issue and revoke hashed API keys per project, with per-key rate limiting baked in.
Usage analytics
Requests, tokens, latency and error rates broken down by day, model and key.
Models
Call any of these with the same request body.
Llama 3.3 70B Versatile
llama-3.3-70b-versatileGeneral-purpose workhorse. Fast, strong at reasoning and long-form text.
GPT-OSS 120B
openai/gpt-oss-120bLarge open-weight model for demanding reasoning and code tasks.
Qwen 3.6 27B (Vision)
qwen/qwen3.6-27bMultimodal: image understanding, OCR, captioning, visual Q&A.
Compound
groq/compoundAgentic system with built-in web search, page visits, and code execution.
Compound Mini
groq/compound-miniSingle tool call per request, ~3x lower latency than Compound.
Get your first key in a minute.
Create an account, generate a key, and start streaming.