---
type: "raw_signal"
id: "muse-spark-open-on-fireworks-meta-s-open-e56ec5"
slug: "muse-spark-open-on-fireworks-meta-s-open-e56ec5"
title: "Muse Spark Open on Fireworks: Meta’s Open Weights Model, Available on US Serverless"
description: "## Fireworks adds Muse Spark Open to US serverless"
retrieval_nugget: "## Fireworks adds Muse Spark Open to US serverless"
observed_at: "2026-10-21"
record_date: "2026-10-21"
date_kind: "observed_at"
why_it_matters: "Source: https://fireworks.ai/blog/meta-muse-spark-open"
novelty: "notable"
verification_level: "source-linked"
signal_type: "release"
source_platform: "fireworks.ai"
topics: []
entities: ["Fireworks AI","Muse Spark Open","Meta","SGLang","Vals AI","Rogo"]
related_patterns: []
novelty_label: "new_fact"
what_changed: "Fireworks now offers Muse Spark Open on US serverless with default Zero Data Retention, benchmark results, and a path to reserved throughput and LoRA fine-tuning."
prior_refs: ["memory:dd504824-bfe3-823d-a2c4-bd6023b73bdc","memory:e5593d4b-a40c-80fa-a02a-4bfa5c680542"]
source_url: "https://fireworks.ai/blog/meta-muse-spark-open"
source_urls: ["https://fireworks.ai/blog/meta-muse-spark-open"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "signal:muse-spark-open-on-fireworks-meta-s-open-e56ec5"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/signals/muse-spark-open-on-fireworks-meta-s-open-e56ec5/","markdown":"https://newruntime.com/signals/muse-spark-open-on-fireworks-meta-s-open-e56ec5.md","json":"https://newruntime.com/signals/muse-spark-open-on-fireworks-meta-s-open-e56ec5.json"}
---

# Muse Spark Open on Fireworks: Meta’s Open Weights Model, Available on US Serverless

## Retrieval answer

## Fireworks adds Muse Spark Open to US serverless

## Fireworks adds Muse Spark Open to US serverless

Fireworks announced that Meta’s open-weight Muse Spark Open is available on its serverless API. The company describes the model as an 816-billion-parameter mixture-of-experts system with 42 billion active parameters per token and a one-million-token context window. It says the endpoint runs on US-only capacity, with no US premium, and that Zero Data Retention is enabled by default. Fireworks also offers reserved throughput for production and serverless LoRA training.

The post reports results from several legal, finance, clinical, and coding evaluations. Those figures are not a uniform independent comparison: the source identifies different evaluators and harnesses, and says some third-party results were measured on Meta’s API release. Fireworks’ announcement establishes platform availability and its stated serving options; it does not independently establish performance for a reader’s workload.

Source: https://fireworks.ai/blog/meta-muse-spark-open

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- collection: [Raw Signals](https://newruntime.com/signals.json): Open every source-linked raw signal.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)