BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

GPT-5.4 vs Muse Spark

Full profiles: GPT-5.4 (OpenAI) and Muse Spark (Meta): every benchmark, price and rank.

Updated · Data from 11 sources

As of 25 Oct 2026, GPT-5.4 (xhigh) and Muse Spark are neck and neck on quality (64.7 vs 64.3, inside the ±4.0 margin of error) and close on price, so pick on the category that matters to you. Pick GPT-5.4 for maths (65 vs 59); pick Muse Spark for instruction following (77 vs 72).

GPT-5.4 benchmarks xhigh OpenAI Muse Spark benchmarks MetaOpen in the comparison tool

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingEither65 vs 65 category score; SWE-Bench Pro 59.1% vs 55.0%Either
Agents & tool useMuse Spark66 vs 64 category score; MCP Atlas 82.2% vs 70.6%Medium
MathsGPT-5.465 vs 59 category score; OTIS Mock AIME 95.3% vs 88.9%Medium
ReasoningGPT-5.469 vs 65 category score; GPQA Diamond 93.3% vs 89.8%Medium
KnowledgeMuse Spark63 vs 60 category score; AA-Omniscience 7.2 vs 5.8Medium
Long documentsGPT-5.466 vs 64 category score; AA-LCR 82.0% vs 78.0%Medium
Images and other mediaMuse Spark65 vs 61 category score; MMMU-Pro 80.5% vs 78.4%Medium
Following instructionsMuse Spark77 vs 72 category score; IFBench 75.9% vs 74.0%Medium
Chat quality (human preference)–Only Muse Spark has an Arena rating (68 category score)Not comparable
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • GPT-5.4 (xhigh) and Muse Spark are neck and neck on quality (64.7 vs 64.3, within the ±4.0 margin).
  • GPT-5.4 (xhigh) is stronger in composite, long context, maths, reasoning.
  • Muse Spark is stronger in agents & tools, coding, instruction following, knowledge, multimodal, human preference.

At a glance

SpecGPT-5.4Muse Spark
DeveloperOpenAIMeta
Released5 Mar 20268 Apr 2026
Context window (tokens)1,050,000262,144
Compared atGPT-5.4 (xhigh)Muse Spark
Open weightsNoNo
Price in / out / cached input per 1M$2.50 / $15.00 / $0.250–

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryGPT-5.4 (xhigh)Muse SparkLeader
Reasoning69.165.3GPT-5.4
Coding64.665.0level
Agents & tools63.666.1Muse Spark
Maths65.259.1GPT-5.4
Knowledge59.763.3Muse Spark
Instruction following72.377.3Muse Spark
Human preference–68.3not comparable
Multimodal61.464.5Muse Spark
Long context66.164.0GPT-5.4
Composite73.164.5GPT-5.4

Where sources disagree

Across the 34 benchmarks both have, GPT-5.4 leads 23, Muse Spark leads 11. Independent sources split: GPT-5.4 leads on AA Intelligence Index v4.3.2; Muse Spark leads on Arena Text.

Overall indexGPT-5.4 (xhigh)Muse SparkLeader
BenchLeader Index64.764.3neck and neck
AA Intelligence Index v4.3.239.031.3GPT-5.4
Arena Text1475†1489Muse Spark

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.

MeasureGPT-5.4 (xhigh)Muse Spark
Blended price per 1M tokens$5.63–
Input / output per 1M$2.50 / $15.00– / –
Cached input per 1M$0.250–
Price per task: chat reply$0.0055–
Price per task: summarize a 30-page report$0.039–
Price per task: code edit$0.037–
Price per task: agentic coding session$0.210–
Price per task: structured extraction$0.0080–
Output speed96 tok/s–
Time to first answer118.0 s–
Time to a complete answer123.2 s–

Each model is compared at its best-ranked reasoning-effort setting. GPT-5.4 (xhigh) scores 64.7 and takes 118 s to its first answer token; GPT-5.4 (low) scores 59.7 in 2 s, GPT-5.4 (high) scores 58.9 in 1 s. Muse Spark scores 64.3. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

116 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherGPT-5.4 (xhigh)Muse Spark
BenchLeader IndexBenchLeader64.764.3
Agents & tools scoreBenchLeader63.666.1
Coding scoreBenchLeader64.665.0
Composite scoreBenchLeader73.164.5
Instruction following scoreBenchLeader72.377.3
Knowledge scoreBenchLeader59.763.3
Long context scoreBenchLeader66.164.0
Maths scoreBenchLeader65.259.1
Multimodal scoreBenchLeader61.464.5
Reasoning scoreBenchLeader69.165.3
Human preference scoreBenchLeader–68.3
Blended price $/MBenchLeader$5.63–
Output speedBenchLeader96 tok/s–
Time to first answerBenchLeader118.0 s–
Context windowBenchLeader1,050,000 tokens262,144 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.3%89.8%
Humanity's Last ExamScale AI / CAIS36.2%40.6%
Arena Hard PromptsArena (LMArena)1487†1506
LiveBench 1.0 ReasoningLiveBench88.1%–
GPQA Diamond (AA)Artificial Analysis92.0%88.4%
Humanity's Last Exam (AA)Artificial Analysis43.7%40.7%
GPQA Diamond (Vals)Vals AI91.7%89.7%
Kagi LLM BenchmarkKagi LLM Benchmark63.8%†–
ARC-AGI-1ARC Prize93.7%–
ARC-AGI-2ARC Prize74.0%–
ARC-AGI-3ARC Prize0.2%†–
CritPtArtificial Analysis23.4%11.3%
Chess PuzzlesEpoch AI Benchmarking Hub48.0%–
Mystery Game PuzzlesEpoch AI Benchmarking Hub37.0%–
CL-benchEpoch AI Benchmarking Hub27.9%–
CL-bench LifeEpoch AI Benchmarking Hub21.7%–
LMCAEpoch AI Benchmarking Hub52.0%–
DTBenchEpoch AI Benchmarking Hub94.4%–
ForecastBenchForecastBench63.1%†–
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub76.9%†–
SciCodeSciCode56.6%51.5%
WeirdMLWeirdML77.7%–
GSO-BenchGSO-Bench31.4%–
Arena CodingArena (LMArena)1511†1529
Arena WebDevArena (LMArena)1401†–
Show all 116 benchmarks
MetricPublisherGPT-5.4 (xhigh)Muse Spark
Coding
LiveBench 1.0 CodingLiveBench77.5%–
LiveCodeBenchVals AI84.1%–
SWE-bench (Vals)Vals AI78.2%74.4%
SWE-Bench ProScale AI SEAL59.1%55.0%
Code MigrationVals AI35.0%–
ProgramBenchVals AI0.0%–
Vibe Code Bench v1.1Vals AI67.4%19.7%
SWE-Bench Pro (private)Scale AI SEAL59.1%55.0%
SWE Atlas: Codebase QnAScale AI SEAL40.8%24.2%
SWE Atlas: RefactoringScale AI SEAL44.3%–
SWE Atlas: Test WritingScale AI SEAL44.4%31.1%
MirrorCodeEpoch AI Benchmarking Hub15.6%†–
DeepSWE v1.1Epoch AI Benchmarking Hub51.8%–
ALE-BenchEpoch AI Benchmarking Hub1607†–
AlgoTuneAlgoTune1.9†–
Terminal-Bench 2.1 (AA)Artificial Analysis78.3%62.2%
Agents & tools
Terminal-BenchTerminal-Bench81.8%†–
APEX-AgentsMercor52.4%†–
LiveBench Agentic CodingLiveBench53.8%–
Terminal-Bench HardArtificial Analysis57.6%45.5%
τ²-Bench Telecom (AA)Artificial Analysis87.1%91.5%
MCP AtlasScale AI SEAL70.6%82.2%
HiL-BenchScale AI SEAL9.7%†–
τ³-bench Bankingτ²-bench39.4%–
τ³-Banking (AA)Artificial Analysis39.6%–
APEX-Agents (AA)Artificial Analysis33.3%–
Harvey's Legal Agent Benchmark v1Vals AI0.0%–
SkillsBenchVals AI51.7%–
Terminal-Bench 2.0 (Vals)Vals AI58.4%†59.5%
DeepResearch BenchDeepResearch Bench35.1%†–
PostTrainBenchEpoch AI Benchmarking Hub19.0%†–
METR Time HorizonsMETR74.3%–
GBAEvalEpoch AI Benchmarking Hub45.1%†–
Vending-Bench 2Andon Labs6144.2†–
DeepSearchQAGoogle63.7%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub78.6%–
FrontierMath Tier 4Epoch AI Benchmarking Hub49.0%–
OTIS Mock AIMEEpoch AI Benchmarking Hub95.3%88.9%
ProofBenchVals AI56.0%17.0%
LiveBench MathematicsLiveBench94.2%–
AIME (Vals)Vals AI96.7%96.9%
AIME 2026MathArena99.2%–
HMMT February 2026MathArena97.7%–
USAMO 2026MathArena95.2%–
MathArena ApexMathArena54.2%–
Arena MathsArena (LMArena)1460†1467
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub45.1%–
LiveBench 1.0 Data AnalysisLiveBench79.3%–
AA-OmniscienceArtificial Analysis5.87.2
MMLU-ProVals AI87.5%87.3%
LegalBenchVals AI86.0%84.2%
CorpFinVals AI65.3%65.1%
TaxEvalVals AI74.0%77.7%
MedQAVals AI96.1%–
PRBench FinanceScale AI SEAL45.6%†52.4%
PRBench LegalScale AI SEAL44.4%†52.3%
MultiNRCScale AI SEAL58.3%†59.0%
CaseLaw v2Vals AI63.8%63.1%
MedCodeVals AI41.3%51.3%
MedScribeVals AI77.5%85.9%
EBR-benchEpoch AI Benchmarking Hub25.4%–
AA-Omniscience: accuracyArtificial Analysis50.9%49.6%
AA-Omniscience: non-hallucinationArtificial Analysis8.3%15.8%
Instruction following
LiveBench LanguageLiveBench82.6%–
LiveBench Instruction FollowingLiveBench70.2%–
IFBenchArtificial Analysis74.0%75.9%
MultiChallengeScale AI SEAL–75.5%
TutorBenchScale AI SEAL–68.5%
Arena Instruction FollowingArena (LMArena)1460†1465
Human preference
Arena TextArena (LMArena)1465†1489
EQ-Bench 4EQ-Bench1260†–
Arena Creative WritingArena (LMArena)1435†1464
Arena Multi-turnArena (LMArena)1480†1494
Arena Longer QueriesArena (LMArena)1477†1476
Multimodal
Arena VisionArena (LMArena)1292†1306
MMMU-ProArtificial Analysis78.4%80.5%
VISTAScale AI SEAL50.9%†–
MMMU-Pro (official)MMMU–80.4%†
MMMU-Pro (Vals)Vals AI87.5%87.4%
MortgageTaxVals AI68.3%–
SAGEVals AI43.3%–
VTBScale AI SEAL29.2%†–
Blueprint-Bench 2Epoch AI Benchmarking Hub27.1%†–
Long context
AA-LCRArtificial Analysis82.0%78.0%
Arena DocumentArena (LMArena)1470†1467
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub157.3†152.0
LiveBench 1.0LiveBench78.0%–
AA Intelligence Index v4.3.2Artificial Analysis39.031.3
Vals RSI IndexVals AI15.5%–
Safety & honesty
FORTRESSScale AI SEAL–20.2%
MASKScale AI SEAL89.7%†–

Data as of 2026-10-25. Best configuration of each model; every score links to its source on the model pages.

What changed

  • GPT-5.4 (xhigh): 2 new results (Harvey's Legal Agent Benchmark v1, Vals RSI Index).
  • GPT-5.4 (xhigh): 2 new results (USAMO 2026, τ³-bench Banking).
  • GPT-5.4 (xhigh): 1 score revision.
  • Muse Spark: 1 new result (PRBench Legal).
  • GPT-5.4 (xhigh): 1 new result (SWE Atlas: Codebase QnA).
  • Muse Spark: 1 new result (SWE Atlas: Codebase QnA).
  • Muse Spark: 2 new results (PRBench Legal, MultiNRC).
  • Muse Spark: 1 new result (PRBench Finance).
  • Muse Spark: 1 new result (PRBench Finance).
  • GPT-5.4 (xhigh): 1 score revision.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

GPT-5.4 vs Muse Spark: questions

Is GPT-5.4 better than Muse Spark?
GPT-5.4 (xhigh) and Muse Spark are neck and neck on quality (64.7 vs 64.3, within the ±4.0 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-25, so check the category scores for your use.
Is GPT-5.4 better than Muse Spark for coding?
They are level in coding (65 vs 65 on the category index). On SWE-Bench Pro, the top shared coding benchmark, it is 59.1% for GPT-5.4 against 55.0% for Muse Spark.
Is GPT-5.4 better than Muse Spark for agents and tool use?
Muse Spark scores higher in agentic tasks (66 vs 64 on the category index, where 50 is average). On MCP Atlas, the top shared agents & tools benchmark, it is 70.6% for GPT-5.4 against 82.2% for Muse Spark.
Which is better for long documents?
GPT-5.4 scores higher on long-context benchmarks (66 vs 64). GPT-5.4 accepts more context: 1,050,000 against 262,144 tokens for Muse Spark.
Do independent benchmarks agree on GPT-5.4 vs Muse Spark?
Across the 34 benchmarks both have, GPT-5.4 leads 23, Muse Spark leads 11. Independent sources split: GPT-5.4 leads on AA Intelligence Index v4.3.2; Muse Spark leads on Arena Text.
When was Muse Spark released?
Muse Spark was released on 8 Apr 2026.