LongCat-2.5 vs Muse Spark 1.3: Specs & Benchmark Comparison

LongCat-2.5 is developed by Meituan, while Muse Spark 1.3 comes from Meta. Muse Spark 1.3 was released in September 2026, and LongCat-2.5 followed 2 months later in November 2026. LongCat-2.5 has a published size of about 1.6 trillion parameters; Meta has not disclosed the parameter count of Muse Spark 1.3.

The two models share 4 published benchmarks. Muse Spark 1.3 leads on 3 of them, LongCat-2.5 on 1. The widest gaps are on DeepSWE 1.1, where Muse Spark 1.3 scores 75.4% against 64.6%; SWE Atlas - Codebase QnA, where Muse Spark 1.3 scores 59.4% against 58.3%. Averaged across everything we track, LongCat-2.5 sits at 68.4% and Muse Spark 1.3 at 73.4%.

LongCat-2.5 is the cheaper API at $0.30 per million input tokens and $1.20 per million output tokens, roughly 4 times cheaper than Muse Spark 1.3 at $1.25 and $4.25. Both accept a context window of about 1M tokens. LongCat-2.5 accepts text, images, and video as input, while Muse Spark 1.3 accepts text, images, audio, and video.

CharacteristicLongCat-2.5Muse Spark 1.3
CompanyMeituanMeta
Release DateNovember 1, 2026September 2, 2026
Parameters1.6T—
MultimodalYesYes
Context (input)1.0M1.0M
Context (output)131K944K
Input Price / 1M$0.30$1.25
Output Price / 1M$1.20$4.25
Average Score68.4%73.4%
Benchmarks
DeepSWE 1.164.6%75.4%
SWE Atlas - Codebase QnA58.3%59.4%
Job Bench65.4%64.9%
Terminal-Bench 2.188.4%88.8%

Visual Benchmark Comparison

LongCat-2.5
Muse Spark 1.3
DeepSWE 1.10.6 vs 0.8
0.6
0.8
SWE Atlas - Codebase QnA0.6 vs 0.6
0.6
0.6
Job Bench0.7 vs 0.6
0.7
0.6
Terminal-Bench 2.10.9 vs 0.9
0.9
0.9

Verdict

LongCat-2.5 leads in 2 out of 4 comparison categories.

Overall Performance

Both models show comparable average scores: LongCat-2.5 — 0.7, Muse Spark 1.3 — 0.7.

API Cost

LongCat-2.5 is 3.7x cheaper: input $0.30/1M vs $1.25/1M tokens.

Context Window

Same context size: 1M tokens.

Recency

LongCat-2.5 is newer: released 11/1/2026 vs 9/2/2026.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — LongCat-2.5 or Muse Spark 1.3?
Direct comparison on the SWE-Bench benchmark is not available. We recommend reviewing other metrics on the comparison page.
Which model is cheaper — LongCat-2.5 or Muse Spark 1.3?
LongCat-2.5 is cheaper for input: $0.30 per 1M tokens vs $1.25.
Which has a larger context window — LongCat-2.5 or Muse Spark 1.3?
LongCat-2.5 supports a larger context: 1,048,576 tokens vs 1,048,576.

The LongCat-2.5 and Muse Spark 1.3 comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the LongCat-2.5 or Muse Spark 1.3 page. See also the complete list of AI model comparisons.