Compare Models

Muse Spark 1.2

Muse Spark 1.2 is a model from Meta, released August 5, 2026. Its best result is #5 of 109 on MedScribe.

Release Date: Aug 5, 2026

Developer Meta  🇺🇸
Context Window 1M
Max Output Tokens 131k
Token Costs (in/out) $1.25/4.25
Weights Private
Input Modalities

Accuracy

55.92 %

Avg. Cost (In/Out)

$ 1.25 / $ 4.25

Latency

16 min 33 s

Vals Index
BenchmarksAccuracyRankings
Code Migration

29.95%

±4.02
43/79
CyberBench v1.1

69.52%

±5.56
23/47
EMB

56.98%

±3.13
43/76
Finance Agent (v2)

60.60%

±0.28
4/81
Immigration

66.54%

±1.59
14/25
Legal Research Bench

43.75%

±3.45
23/90
MedCode

49.35%

±2.19
24/107
MedScribe

90.06%

±1.96
5/109
MortgageTax

65.42%

±0.93
43/98
ProofBench v1.1

43.00%

±4.98
34/52
Public Benefits Bench v1.1

68.47%

±1.21
11/50
SRE Bench

0.00%

±0.00
34/36
Tax Agent Bench

32.80%

±2.75
19/81
TaxEval v2

80.38%

±0.76
1/145
Vibe Code Bench v1.1

79.10%

±3.31
31/124
BioMysteryBench

64.81%

±1.33
22/29
IOI

21.78%

±0.87
43/46
LegalBench

85.26%

±0.45
23/151
MMLU Pro

88.28%

±0.32
20/138
MMMU Pro

86.13%

±0.83
22/93
ProgramBench

0.50%

±0.50
29/64
SkillsBench

53.04%

±4.42
17/35
SWE-bench

86.60%

±1.52
14/88
Terminal-Bench 4.0

6.06%

±1.51
50/60
Proprietary BenchmarksAcademic BenchmarksIndustry Partners
Vals
Default Provider : Meta
Temperature: 1
Top P: Default
Top K: Default
Max Output Tokens: 131,072
Reasoning Effort: xhigh

Compared with

Muse Spark 1.2 vs Step 5 Preview 9 shared benchmarks Muse Spark 1.2 vs Mistral Large 4 Preview 10 shared benchmarks Muse Spark 1.2 vs Hy4 Preview 14 shared benchmarks Muse Spark 1.2 vs Kimi K3 19 shared benchmarks Muse Spark 1.2 vs DeepSeek V4 Flash 0731 16 shared benchmarks Muse Spark 1.2 vs Qwen 3.8 Max 19 shared benchmarks Muse Spark 1.2 vs DeepSeek V4 Pro 0813 17 shared benchmarks Muse Spark 1.2 vs Ember-1 9 shared benchmarks Muse Spark 1.2 vs Gemini 3.5 Flash 16 shared benchmarks Muse Spark 1.2 vs GPT-6 Luna 14 shared benchmarks