Muse Spark 1.2 Contributor vs Muse Spark 1.2

Short answer: two model ids, one checkpoint

LLMTR (llmtr.com) is a gateway that gives access to language models hosted in Türkiye and at global providers through one OpenAI-compatible API. For this comparison, send POST requests to https://llmtr.com/v1/chat/completions through the OpenAI-compatible base URL https://llmtr.com/v1. The model strings are meta/muse-spark-1.2-contributor and meta/muse-spark-1.2.

Choose the standard model for personal, customer, confidential or regulated data because its prompts and completions are not used by Meta for training. Choose Contributor for public content, synthetic data, open-source code, evaluation and test prompts; Meta uses its prompts and completions for training, and it has a lower rate limit.

Muse Spark 1.2 Contributor vs Muse Spark 1.2

They are two LLMTR listings for the same Muse Spark 1.2 checkpoint, not different model versions or separate capability sets under one OpenAI-compatible API. As of the 2026-10-06 catalog, meta/muse-spark-1.2 costs $1.25 per 1M input tokens, while meta/muse-spark-1.2-contributor costs $0.10.

The trade-off is price, training-data use and rate limit. Contributor is 12.5 times cheaper for input, 21.25 times cheaper for output and 75 times cheaper for cache reads. Meta uses Contributor prompts and completions to train future models, and Contributor has a lower rate limit; standard prompts and completions are not used for training.

What is the same: checkpoint, features and context

Both model ids accept text, image, video and document input. They also support function calling, JSON mode, structured outputs and prompt caching. These capabilities are identical because both listings serve the same Muse Spark 1.2 checkpoint. They expose the same feature set through the same API.

Both provide a 1,048,576-token context window and are served only on POST https://llmtr.com/v1/chat/completions. Reasoning is always on for both, with no none setting. The reasoning_effort parameter accepts minimal, low, medium, high and xhigh, and reasoning tokens are counted in completion_tokens.

• Meta, Muse Spark models · 2026-10-06

What is different and what it costs

Contributor input is 12.5 times cheaper, output is 21.25 times cheaper and cache reads are 75 times cheaper than standard. The 2026-10-06 catalog prices per 1M tokens are $1.25 input, $4.25 output and $0.15 cache read for standard, versus $0.10, $0.20 and $0.002 for Contributor. Reasoning tokens count in completion_tokens and are billed at the output price on both.

Data use and traffic handling differ. Meta uses Contributor prompts and completions to train future models; standard prompts and completions are not used for training. Contributor has a lower rate limit, so requests under heavy traffic can receive 429 and need retry with backoff. The table assumes 10M input and 2M output tokens monthly.

Which Muse Spark 1.2 tier to choose

Use meta/muse-spark-1.2 for personal data, customer data, confidential code or documents, and regulated data. Use meta/muse-spark-1.2-contributor for public content, synthetic data, open-source code, evaluation prompts and test prompts. This rule accounts for the Contributor training-data use and its lower rate limit.

The routing code below changes the model string according to the data classification. It leaves the OpenAI client, base URL https://llmtr.com/v1, API key, messages and reasoning_effort unchanged. Switching later also changes only the model string; the same API key and credit balance apply.

Select a Muse Spark 1.2 model id based on whether the request contains private data.

import os

from openai import OpenAI

client = OpenAI(base_url="https://llmtr.com/v1", api_key=os.environ["LLMTR_API_KEY"])


def pick_model(contains_private_data: bool) -> str:
    # Contributor prompts and outputs are used by Meta for training.
    if contains_private_data:
        return "meta/muse-spark-1.2"
    return "meta/muse-spark-1.2-contributor"


response = client.chat.completions.create(
    model=pick_model(contains_private_data=False),
    messages=[{"role": "user", "content": "Summarize this public changelog in three bullets: ..."}],
    reasoning_effort="low",
)
print(response.choices[0].message.content)

Getting started on LLMTR: account, credit, key and first request

Step 1: create an account at llmtr.com and verify your email. Step 2: top up credit from the Dashboard through secure checkout. An 8% platform margin is added once: a $10.00 credit top-up is charged at $10.80. This margin is not added to model prices; calls are deducted at catalog prices. Step 3: create an API key under Dashboard > API Keys. The raw key is shown only once, so store it as LLMTR_API_KEY.

Step 4: send the first request with any OpenAI-compatible SDK by setting base_url to https://llmtr.com/v1 and api_key to the stored key. The request uses POST https://llmtr.com/v1/chat/completions. Usage and spend per key appear on the Dashboard usage page. The same key and credit balance work across the catalog; switching models changes only the model string.

Set up LLMTR and call Muse Spark 1.2

Create an account, add credit, create an API key and send a request through the OpenAI-compatible endpoint.

1. Create an account. Create an account at llmtr.com and verify your email.
2. Top up credit. Top up credit from the Dashboard through secure checkout. The requested amount is charged once with an 8% platform margin, so $10.00 credit costs $10.80; model prices do not include this margin.
3. Create an API key. Create a key under Dashboard > API Keys. Copy the raw key when it is shown once and save it as LLMTR_API_KEY or another environment variable.
4. Send the first request. Configure any OpenAI-compatible SDK with base_url https://llmtr.com/v1 and api_key from the environment, then send a POST request to https://llmtr.com/v1/chat/completions with a catalog model id.

Frequently asked questions

Is Muse Spark 1.2 Contributor a different model?

No. Both use the same Muse Spark 1.2 checkpoint and 1,048,576-token context. Contributor costs $0.10 per 1M input tokens versus $1.25, uses prompts and completions for training, and has a lower rate limit; standard does not use prompts or completions for training.

Can I send customer data to Muse Spark 1.2 Contributor?

Do not send customer data to meta/muse-spark-1.2-contributor. Use meta/muse-spark-1.2, whose prompts and completions are not used by Meta for training. Contributor prompts and completions are used to train future models.

Is there a Muse Spark 1.3 version on LLMTR?

Yes. LLMTR lists meta/muse-spark-1.3 and meta/muse-spark-1.3-contributor. Their input, output and cache-read prices match the 1.2 tiers: standard at $1.25, $4.25 and $0.15 per 1M tokens; Contributor at $0.10, $0.20 and $0.002.

Related posts

• The real difference between Muse Spark Contributor and the standard version: not quality, price and data policy
• What is Muse Spark 1.2 Contributor? Data policy and the standard tier
• Muse Spark 1.2 Contributor cost: cache and reasoning accounting

• Muse Spark 1.2 Contributor model
• Muse Spark 1.2 model
• Pricing