LLM providers

Availability

Open core · self-host + all Splyntra Cloud plans

LLM calls become llm_call spans with model name, token counts, and cost. Splyntra has three native provider instrumentors, and captures any OpenAI-compatible provider through the OpenAI instrumentor.

Native instrumentors

Provider	instrument name	Languages
OpenAI	openai	Python + TS
Anthropic (Claude)	anthropic	Python + TS
Ollama	ollama	Python + TS

from splyntra import Splyntra

Splyntra(
    api_key=os.environ["SPLYNTRA_API_KEY"],
    project="my-app",
    instrument=("openai", "anthropic"),
)

OpenAI-compatible providers

Many providers expose an OpenAI-compatible API. Point your OpenAI client's base_url at the provider and keep instrument=("openai",) — the OpenAI adapter captures the calls and Splyntra labels and cost-tracks them by provider.

Provider	base_url
Grok / xAI	https://api.x.ai/v1
Google Gemini	https://generativelanguage.googleapis.com/v1beta/openai/
Groq	https://api.groq.com/openai/v1
OpenRouter	https://openrouter.ai/api/v1
Mistral	provider's OpenAI-compatible endpoint
Together	provider's OpenAI-compatible endpoint
Fireworks	provider's OpenAI-compatible endpoint
DeepSeek	provider's OpenAI-compatible endpoint
Cohere	provider's OpenAI-compatible endpoint

Example — Grok/xAI via the OpenAI client

Python

import os
from splyntra import Splyntra
from openai import OpenAI

Splyntra(
    api_key=os.environ["SPLYNTRA_API_KEY"],
    project="my-app",
    instrument=("openai",),   # keep the OpenAI instrumentor
)

client = OpenAI(
    api_key=os.environ["XAI_API_KEY"],
    base_url="https://api.x.ai/v1",   # point at the provider
)
client.chat.completions.create(
    model="grok-2-latest",
    messages=[{"role": "user", "content": "Hello"}],
)

TypeScript

import { Splyntra } from "@splyntra/sdk";
import OpenAI from "openai";

new Splyntra({
  apiKey: process.env.SPLYNTRA_API_KEY!,
  project: "my-app",
  instrument: ["openai"], // keep the OpenAI instrumentor
});

const client = new OpenAI({
  apiKey: process.env.XAI_API_KEY!,
  baseURL: "https://api.x.ai/v1", // point at the provider
});
await client.chat.completions.create({
  model: "grok-2-latest",
  messages: [{ role: "user", content: "Hello" }],
});

Swap the base_url and model for any provider in the table — the instrumentation is identical.

tip

Cost attribution uses your model pricing. If a provider or model isn't priced yet, add it under Costs → Manage model pricing. See Costs.

Next steps

• Framework integrations — combine a provider with LangGraph, CrewAI, and more.
• Costs — spend by model and provider.
• Connect an agent — the wizard emits the base_url notes for you.