Python

pip install openai

from openai import OpenAI
client = OpenAI(
    api_key="nim_live_...",
    base_url="https://meridian.getnimbus.net/api/v1",
)

# Let Meridian pick the best model per request.
# See /docs/router for tags + priorities.
r = client.chat.completions.create(
    model="auto",
    messages=[
        {"role": "system", "content": "You are a senior engineer."},
        {"role": "user",   "content": "Refactor this Python class for clarity ..."},
    ],
)
print(r.choices[0].message.content)

from openai import OpenAI
client = OpenAI(
    api_key="nim_live_...",
    base_url="https://meridian.getnimbus.net/api/v1",
)
r = client.chat.completions.create(
    model="azure/gpt-5-mini",
    messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")
with client.chat.completions.stream(
    model="azure/gpt-5-mini",
    messages=[{"role": "user", "content": "Stream a haiku about code."}],
) as stream:
    for event in stream:
        if event.type == "token":
            print(event.token, end="", flush=True)
    print()

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")
res = client.embeddings.create(
    model="azure/text-embedding-3-large",
    input=["Hello world", "Goodbye world"],
)
print(len(res.data), len(res.data[0].embedding))

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")

with open("speech.mp3", "rb") as f:
    resp = client.audio.transcriptions.create(model="whisper-1", file=f)
print(resp.text)

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")

resp = client.audio.speech.create(model="tts-1", voice="alloy", input="Hello from Meridian")
resp.stream_to_file("out.mp3")

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")

r = client.responses.create(
    model="azure/gpt-5-mini",
    input=[
        {"role": "system", "content": "You are concise."},
        {"role": "user", "content": "Write a tagline for a developer SaaS."},
    ],
    modalities=["text"],
)
print(r.output_text)

# Cohere-compatible rerank shape via Meridian
import requests

headers = {"Authorization": "Bearer nim_live_...", "Content-Type": "application/json"}
body = {
    "model": "nimbus/rerank-1",
    "query": "best snack for hiking",
    "documents": [
        "Trail mix with nuts and dried fruit",
        "Chocolate cake recipe",
        "Electrolyte drink comparison",
    ],
    "top_n": 2,
}
resp = requests.post("https://meridian.getnimbus.net/api/v1/rerank", headers=headers, json=body, timeout=30)
print(resp.json())

# Upload a file
import requests
headers = {"Authorization": "Bearer nim_live_..."}
with open("policy.pdf", "rb") as f:
    r = requests.post(
        "https://meridian.getnimbus.net/api/v1/files",
        headers=headers,
        files={"file": ("policy.pdf", f)},
        data={"purpose": "assistants"},
        timeout=60,
    )
print(r.json())

# Download a file by id
import requests
headers = {"Authorization": "Bearer nim_live_..."}
fid = "<file_id_or_url>"
r = requests.get(f"https://meridian.getnimbus.net/api/v1/files?id={fid}", headers=headers, timeout=60)
open("download.bin", "wb").write(r.content)

# Delete a file by id
import requests
headers = {"Authorization": "Bearer nim_live_..."}
fid = "<file_id_or_url>"
r = requests.delete(f"https://meridian.getnimbus.net/api/v1/files?id={fid}", headers=headers, timeout=30)
print(r.json())

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")

img = client.images.generate(
    model="azure/gpt-image-1",
    prompt="Violet and emerald near-black landing page hero, minimalist",
    size="1024x1024",
    n=1,
)
import base64
img_b64 = img.data[0].b64_json
open("image.png", "wb").write(base64.b64decode(img_b64))

from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")
msgs = [
    {"role": "system", "content": "You are concise."},
    {"role": "user", "content": "Summarize in 10 words: Python generators."},
]
first = client.chat.completions.create(model="azure/gpt-5-mini", messages=msgs)
msgs.append({"role": "assistant", "content": first.choices[0].message.content})
msgs.append({"role": "user", "content": "Now add an example."})
second = client.chat.completions.create(model="azure/gpt-5-mini", messages=msgs)
print(second.choices[0].message.content)

from openai import OpenAI
import time
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")

def call_with_retry():
    backoff = 0.5
    for attempt in range(6):
        try:
            return client.chat.completions.create(
                model="azure/gpt-5-mini",
                messages=[{"role": "user", "content": "Hello"}],
            )
        except Exception as e:
            msg = str(e)
            if any(code in msg for code in ["429", "503"]):
                time.sleep(backoff)
                backoff = min(8.0, backoff * 2)
                continue
            if "401" in msg:
                raise RuntimeError("Check API key format nim_live_... and balance")
            raise

resp = call_with_retry()
print(resp.choices[0].message.content)

# raw HTTP balance check
import requests
headers = {"Authorization": "Bearer nim_live_..."}
print(requests.get("https://meridian.getnimbus.net/api/v1/balance", headers=headers, timeout=15).json())

# Every chat / embedding response carries Meridian billing fields.
# Two equivalent ways to read them.
from openai import OpenAI
client = OpenAI(api_key="nim_live_...", base_url="https://meridian.getnimbus.net/api/v1")
r = client.chat.completions.create(
    model="azure/gpt-5-mini",
    messages=[{"role": "user", "content": "Hello"}],
)

# (1) Inline body fields (additive to OpenAI usage block).
raw = r.model_dump()
usage = raw["usage"]
print("user_tokens spent     :", usage["user_tokens"])
print("balance after request :", usage["balance_after_user_tokens"])
print("provider tokens       :", usage["prompt_tokens"], usage["completion_tokens"])

# (2) Custom response headers (use when you do not want to deserialize the body).
with client.with_raw_response.chat.completions.create(
    model="azure/gpt-5-mini",
    messages=[{"role": "user", "content": "Hello"}],
) as raw_resp:
    print("X-Meridian-Tokens-Used   :", raw_resp.headers.get("x-meridian-tokens-used"))
    print("X-Meridian-Balance-After :", raw_resp.headers.get("x-meridian-balance-after"))