Muse Spark 1.3
Meta’s closed flagship for long coding sessions, with near-perfect recall across a million tokens.
About Muse Spark 1.3
Muse Spark is Meta’s proprietary line from Meta Superintelligence Labs — unlike Llama, its weights are not released. Version 1.3, from September 2026, is built for long agent sessions and coding across large repositories; Meta reports 75.4% on DeepSWE and 88.8% on Terminal-Bench 2.1.
Its most unusual strength is recall over long inputs: 98.1% on the MRCR test between half a million and a million tokens, so it finds what you need even at the far end of a huge document. Meta also says it is harder to trick with injected instructions, and more careful about actions that cannot be undone.
Where it’s strong
Long coding agents across large repositories.
Finding specific facts in very long documents.
Multi-step agent and computer-use tasks.
Where to be careful
Meta’s claim of 25% fewer tokens than version 1.2 did not hold in one independent test, where cost rose 12%.
Reportedly weaker at understanding audio than the previous version.
Closed weights: unlike Llama, you cannot run it yourself.
Price
For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $10.40 — a long prompt and a page of answer each. Every rate is published and stays the same from one request to the next. See every model.
Specs
muse-spark-1.3Call it
Use the OpenAI SDK you already have — point it at Roar AI and name the model. Read the docs.
from openai import OpenAI
client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")
reply = client.chat.completions.create(
model="muse-spark-1.3",
messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)Compare Muse Spark 1.3
Questions
How much does Muse Spark 1.3 cost?
Input is $1.625 per 1M tokens; output is $5.525 per 1M tokens. For scale, 1,000 requests of about 3,000 tokens in and 1,000 out cost $10.40. The rate is published and does not change from one request to the next.
Can I use Muse Spark 1.3 alongside other models?
Yes. Every model on Roar AI is on the same API key and the same monthly invoice, so switching from Muse Spark 1.3 to another model is a change to one word in your code.
Is Muse Spark 1.3 open source?
No. Meta does not publish the weights, so it is only available through an API like this one.