← All models
NVIDIA: Nemotron 3 Ultra
nvidia/nemotron-3-ultra-550b-a55btextReasoning
NVIDIA open frontier model — 550B parameters with 55B active (hybrid Mamba-Transformer MoE). Strong at reasoning and agent orchestration, adjustable reasoning depth, 512k context.
Price in / out
Cache, read / write
Context / Max output
203K / 183K
Released
Jun 4, 2026
Additional rates
Flex, in / out
Parameters
messagesrequiredreasoning_effortoff · medium · highmax_tokensoptional · alias max_completion_tokensstreamtrue · falsetoolsfunction calling (OpenAI format)tool_choiceauto · none · requiredresponse_formatjson_object · json_schemaweb_search · · · providerrouting: order · only · ignore · sortservice_tierflexstream_options{"include_usage": true}Fine-tuning · 9
temperature0–2stopoptionalseedoptionaltop_p0–1top_koptionalmin_p0–1frequency_penalty−2…2presence_penalty−2…2repetition_penalty0–2How to use via API
from openai import OpenAI
client = OpenAI(base_url="https://api.mixen.ai/v1", api_key="mxn-...")
resp = client.chat.completions.create(
model="nvidia/nemotron-3-ultra-550b-a55b",
messages=[{"role": "user", "content": "Привет"}],
reasoning_effort="medium", # off · medium · high
temperature=0.7,
max_tokens=1024,
stream=False,
)
print(resp.choices[0].message.content)
# Стриминг: stream=True — ответ приходит чанками
for chunk in client.chat.completions.create(
model="nvidia/nemotron-3-ultra-550b-a55b",
messages=[{"role": "user", "content": "Привет"}],
stream=True,
):
print(chunk.choices[0].delta.content or "", end="", flush=True)
# эконом-тир: тот же ответ вдвое дешевле, но ниже приоритет в очереди
cheap = client.chat.completions.create(
model="nvidia/nemotron-3-ultra-550b-a55b",
messages=[{"role": "user", "content": "Привет"}],
extra_body={"service_tier": "flex"},
)
