← Blog

Meta Llama 3.3 70B — OpenAI-compatible API

How to call Llama 3.3 70B using any OpenAI-compatible SDK. Hosted on NVIDIA NIM through InferAll at the open-model rate.

InferAll Team

2 min read
Llama 3.3Meta AILLM APINVIDIA NIMOpenAI APIopen source

Meta's Llama 3.3 70B (meta/llama-3.3-70b-instruct) is available via NVIDIA NIM through InferAll at our open-model rate — the cheapest tier in the gateway. Llama 3.3 70B is the refined final iteration of the Llama 3.x 70B line — more instruction-following polish and better benchmark performance than 3.1 70B, at the same model size.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.inferall.ai/v1",
    api_key="ifu_your_key_here",  # get one at inferall.ai/keys
)

response = client.chat.completions.create(
    model="meta/llama-3.3-70b-instruct",
    messages=[{"role": "user", "content": "Explain the difference between Llama 3.1, 3.3, and 4."}],
    max_tokens=400,
)

print(response.choices[0].message.content)

Llama 3.3 70B vs 3.1 70B vs Llama 4

Llama 3.1 70B (meta/llama-3.1-70b-instruct) — the original stable 70B model. Widely tested, very reliable baseline.

Llama 3.3 70B (meta/llama-3.3-70b-instruct) — the refined version. Better instruction following, improved math and reasoning, same 70B architecture. Use this when you need Llama 3.x reliability with better task performance.

Llama 3.1 70B (meta/llama-3.1-70b-instruct) — the previous generation of the same 70B line. Slightly weaker instruction following than 3.3, but a stable baseline that has been serving unchanged for months.

All three are open models on NVIDIA NIM through InferAll, at the same open-model rate.


TypeScript / Node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.inferall.ai/v1",
  apiKey: process.env.INFERALL_API_KEY,
});

const response = await client.chat.completions.create({
  model: "meta/llama-3.3-70b-instruct",
  messages: [{ role: "user", content: "Summarize the key differences between REST and GraphQL." }],
});

Streaming

with client.chat.completions.create(
    model="meta/llama-3.3-70b-instruct",
    messages=[{"role": "user", "content": "Write a guide to async programming in Python."}],
    stream=True,
) as stream:
    for chunk in stream:
        print(chunk.choices[0].delta.content or "", end="")

Claude Code / Cline / Cursor

export ANTHROPIC_BASE_URL=https://api.inferall.ai
export ANTHROPIC_API_KEY=ifu_your_key_here

Llama 3.3 70B serves as the "sonnet-tier" model for Anthropic-compatible clients — balanced performance at the open-model rate.


Llama models on InferAll

Model Size Notes
meta/llama-3.3-70b-instruct 70B Best Llama 3.x, refined instruction following
meta/llama-3.1-70b-instruct 70B Original 70B baseline
meta/llama-3.1-8b-instruct 8B Fast, lightweight
meta/llama-3.2-90b-vision-instruct 90B Vision + text

All on NVIDIA NIM at our open-model rate.


Get started

Sign up at inferall.ai/keys and fund a key with the $5 starter pack — that $5 becomes usage credit you can spend on any model (Llama, GPT, Claude, Gemini, NIM open models) at the provider's published rate with zero markup.