# Muse Spark Open Usage, Cost & Rank | OpenCode Data

> Muse Spark Open costs $1.25 per 1M input tokens and $4.25 per 1M output tokens.

Muse Spark Open is an 816B total / 42B active open-weight multimodal reasoning model from Meta for coding and long-horizon agentic work, with 1M-token context.

- Page: https://opencode.ai/data/meta/muse-spark-open
- JSON: https://opencode.ai/data/meta/muse-spark-open.json
- Model ID: meta/muse-spark-open
- Lab: Meta

## Model facts

| Fact | Value |
| --- | --- |
| Context window | 1M |
| Max output | 131K |
| Knowledge cutoff | 2026-01-04 |
| Release date | 2026-10-15 |
| Input modalities | text, image, video, audio |
| Output modalities | text |
| Reasoning | Yes |
| Tool calling | Yes |
| Open weights | Yes |
| Weights | [Hugging Face (BF16)](https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B) |
| Weights | [Hugging Face (NVFP4)](https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B-NVFP4) |

## Pricing: USD per 1M tokens

| Input | Output | Cached input | Cache write |
| --- | --- | --- | --- |
| $1.25 | $4.25 | $0.15 | - |

## Benchmarks

| Benchmark | Score | Metric | Source |
| --- | --- | --- | --- |
| Artificial Analysis Intelligence Index | 39 | index score | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| JobBench | 61.8 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| MCP Atlas | 89.2 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| GDPval-AA | 1401 | Elo | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| OSWorld-Verified | 80.4 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| Toolathlon-Verified | 72.8 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| HLE-Diamond | 26.1 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| AA-Omniscience | 25 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| SciCode | 58.5 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| Humanity's Last Exam | 61 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| GPQA Diamond | 89 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| DeepSWE | 60 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| SWE-Atlas Codebase QnA | 58 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| Terminal-Bench | 76.4 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| ZeroBench | 50 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| BabyVision | 80.2 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| SimpleVQA | 73.6 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| WorldVQA | 49.4 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| IntPhys2 | 68.3 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |
| Video-MME | 61.5 |  | https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B |

## Methodology

- Updates: Aggregated every hour. Days and weeks use UTC.
- Tokens: Input, output, reasoning, and cached tokens for each request.
- Users and sessions: Approximate counts of distinct users and OpenCode sessions.
- Cost: Session cost is the average cost per OpenCode session. Token prices are list prices from the OpenCode model catalog.
- Retention: The share of a model's users in one week who use it again the next week.
- Citation: Cite OpenCode Data (opencode.ai/data) with the update time shown at the top of the page.