---
format: "aidr-story-markdown/v1"
id: "096d9e30001cb1d14ac020e027846976e36e8ebc5f66877f5d7ac34ab27c472b"
canonical_url: "https://aidr.today/096d9e30?lang=en"
title: "Meta Muse Spark 1.3 Hits 88.8 Coding Score to Match GPT-5.6 Sol"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-02T20:54:48.000Z"
category: "Models"
topics: ["meta","coding","reasoning","benchmark","llm"]
source_urls: ["https://huggingnews.com/ai/update-meta-muse-spark-13-hits-888-coding-score-to-match-gpt-56-sol-1555bd83","https://x.com/finkd/status/2095232032896946311","https://x.com/alexandr_wang/status/2095232919229096121","https://x.com/AIatMeta/status/2095234385129963666","https://x.com/AIatMeta/status/2095234387273269349","https://x.com/wallstengine/status/2095233180425159136","https://x.com/levie/status/2095234253613359200","https://x.com/business/status/2095235349865693194"]
summary: "Meta deployed its latest artificial intelligence model via the Meta Model API and Muse Code on Sept. 2 to increase agentic and coding capabilities. The model achieved a score of 88.8 on Terminal-Bench 2.1, matching GPT-5.6 Sol, and recorded 75.4 on DeepSWE. Internal comparisons show the update utilizes 20% fewer tool calls and 25% fewer tokens than the Muse Spark 1.2 version. The update sustains long-horizon work across multiple workflows in a single thread and is calibrated to avoid hallucinating outcomes. CEO Mark Zuckerberg announced that a max reasoning mode will be released after safety testing, followed by a model named Watermelon and open weights for the Muse Spark line. The model is currently accessible through the Meta Model API, Muse Code, and OpenRouter."
---

# Meta Muse Spark 1\.3 Hits 88\.8 Coding Score to Match GPT\-5\.6 Sol

> [Open the canonical story](<https://aidr.today/096d9e30?lang=en>)

**Published:** 2026-09-02T20:54:48.000Z
**Category:** Models
**Topics:** meta, coding, reasoning, benchmark, llm

## Summary

Meta deployed its latest artificial intelligence model via the Meta Model API and Muse Code on Sept\. 2 to increase agentic and coding capabilities\. The model achieved a score of 88\.8 on Terminal\-Bench 2\.1, matching GPT\-5\.6 Sol, and recorded 75\.4 on DeepSWE\. Internal comparisons show the update utilizes 20% fewer tool calls and 25% fewer tokens than the Muse Spark 1\.2 version\. The update sustains long\-horizon work across multiple workflows in a single thread and is calibrated to avoid hallucinating outcomes\. CEO Mark Zuckerberg announced that a max reasoning mode will be released after safety testing, followed by a model named Watermelon and open weights for the Muse Spark line\. The model is currently accessible through the Meta Model API, Muse Code, and OpenRouter\.

## Sources

- [Story source](<https://huggingnews.com/ai/update-meta-muse-spark-13-hits-888-coding-score-to-match-gpt-56-sol-1555bd83>)
- [Story source](<https://x.com/finkd/status/2095232032896946311>)
- [Supporting source](<https://x.com/alexandr_wang/status/2095232919229096121>)
- [Supporting source](<https://x.com/AIatMeta/status/2095234385129963666>)
- [Supporting source](<https://x.com/AIatMeta/status/2095234387273269349>)
- [Supporting source](<https://x.com/wallstengine/status/2095233180425159136>)
- [Supporting source](<https://x.com/levie/status/2095234253613359200>)
- [Supporting source](<https://x.com/business/status/2095235349865693194>)