MetaReasoning modelCompare with another model

Muse Spark 1.1

Reasoning effort

Muse Spark 1.1 is a Meta proprietary reasoning model, released 9 Jul 2026. Its best configuration ranks #75 of 775 on the BenchLeader Index at 62.9 ±5.7, in the upper half. The ± is the point: 186 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (74) and lowest in agents & tools (51). At $2.00 per million tokens blended it is pricier than most ranked models. Output speed of 236 tokens per second puts it in the fastest quarter, with a first token in 1.7 s. It has been measured at 4 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 19 Oct 2026.

Blended price: $2.00/M; $1.25 in · $4.25 out
Output speed: 236 tok/s; OpenRouter traffic, 7-day median; not yet measured by Artificial Analysis
First answer: 1.68 s
Context: 1.0M
Full answer: –
Cost per run: –
Released: 9 Jul 2026

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index63
Reasoning68
Coding66
Agents & tools51
Maths62
Knowledge69
Instruction following74
Human preference66
Multimodal64

Versions

Meta has shipped 4 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

Model	Released	Index	Rank
Muse Spark 1.3max	2 Sept 2026	67.2	#26
Muse Spark 1.2xhigh	5 Aug 2026	63.2	#69
Muse Spark 1.1this page	9 Jul 2026	62.9	#75
Muse Spark	8 Apr 2026	64.4	#50

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
medium	–	–	236 tok/s	1.68 s	$0.0018
high	–	–	236 tok/s	1.68 s	$0.0018
xhigh	61.6	#95	280 tok/s	21 s	$0.0018	Agents & tools 59 · Coding 64 · Composite 67 · Knowledge 67 · Long context 64 · Reasoning 66
not statedbest	62.9	#75	236 tok/s	1.68 s	$0.0018	Agents & tools 51 · Coding 66 · Human preference 66 · Instruction following 74 · Knowledge 69 · Maths 62 · Multimodal 64 · Reasoning 68

How its index has moved

8 Sept 2026 to 20 Oct 2026

not stated
xhigh

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	high	xhigh	not stated	Source
Arena Hard Promptseffort not stated	–	–	1511#18	Arena (LMArena)
LiveBench 1.0 Reasoningnot in index	–	87.7%#26	–	LiveBench
GPQA Diamond (AA)not in index	–	89.8%#68	–	Artificial Analysis
Humanity's Last Exam (AA)not in index	–	46.2%#45	–	Artificial Analysis
GPQA Diamond (Vals)not in index	–	91.2%#23	–	Vals AI
CritPt	–	15.1%#92	–	Artificial Analysis
Chess Puzzlesnot in index	–	34.0%#57	–	Epoch AI Benchmarking Hub
LMCAnot in index	49.9%#65	–	–	Epoch AI Benchmarking Hub
DTBenchnot in index	94.4%#56	–	–	Epoch AI Benchmarking Hub

Coding

Benchmark	high	xhigh	not stated	Source
SciCode	–	58.8%#17	58.8%#17	SciCode
Arena Codingeffort not stated	–	–	1533#19	Arena (LMArena)
Arena WebDeveffort not stated	–	–	1542#43	Arena (LMArena)
LiveBench 1.0 Codingnot in index	–	77.2%#39	–	LiveBench
SciCode (AA)not in index	–	58.8%#18	–	Artificial Analysis
LiveCodeBench	–	85.9%#27	–	Vals AI
SWE-bench (Vals)not in index	–	82.0%#21	–	Vals AI
SWE-Bench Proeffort not stated	–	–	61.5%#1	Scale AI SEAL
Code Migrationnot in index	–	31.1%#39	–	Vals AI
ProgramBenchnot in index	–	0.0%#32	–	Vals AI
Vibe Code Bench v1.1not in index	–	72.2%#38	–	Vals AI
SWE-Bench Pro (private)not in indexeffort not stated	–	–	61.5%#1	Scale AI SEAL
SWE Atlas: Codebase QnAnot in index	–	42.2%#9	–	Scale AI SEAL
SWE Atlas: Test Writingnot in index	–	41.5%#10	–	Scale AI SEAL
Surface Evolver Benchnot in index	52.5%#17	–	–	Epoch AI Benchmarking Hub
DeepSWE v1.1not in indexeffort not stated	–	–	53.3%#45	Epoch AI Benchmarking Hub
Terminal-Bench 4.0 (AA)not in index	–	6.1%#97	–	Artificial Analysis
Terminal-Bench 2.1 (AA)not in index	–	77.9%#54	–	Artificial Analysis

Agents & tools

Benchmark	xhigh	not stated	Source
APEX-Agentseffort not stated	–	31.8%#48	Mercor
Arena Agenteffort not stated	–	-6.6#42	Arena (LMArena)
LiveBench Agentic Codingnot in index	58.5%#18	–	LiveBench
Terminal-Bench 2.1 (Vals)	69.3%#30	–	Vals AI
MCP Atlaseffort not stated	–	88.1%#1	Scale AI SEAL
τ³-bench Bankingnot in index	40.5%#6	–	τ²-bench
GDPval-AA v2.1not in index	35.6%#117	–	Artificial Analysis
τ³-Banking (AA)	31.8%#62	–	Artificial Analysis
Finance Agent v2not in index	57.2%#16	–	Vals AI
Legal Research Benchnot in index	38.0%#37	–	Vals AI
SkillsBenchnot in index	59.2%#10	–	Vals AI
Time Horizon Index: KSPnot in index	1.7%#13	–	Vals AI
GBAEvalnot in indexeffort not stated	–	7.9%#13	Epoch AI Benchmarking Hub
Vending-Bench 2not in indexeffort not stated	–	6520.5#18	Andon Labs
AutomationBenchnot in index	38.9%#99	–	Artificial Analysis
GDP.pdf (AA)not in index	14.4%#100	–	Artificial Analysis

Maths

Benchmark	xhigh	not stated	Source
ProofBencheffort not stated	–	39.0%#38	Vals AI
LiveBench Mathematicsnot in index	87.1%#47	–	LiveBench
Arena Mathseffort not stated	–	1496#20	Arena (LMArena)

Knowledge

Benchmark	medium	xhigh	not stated	Source
SimpleQA Verifiedeffort not stated	–	–	57.8%#18	Epoch AI Benchmarking Hub
LiveBench 1.0 Data Analysisnot in index	–	72.5%#45	–	LiveBench
AA-Omniscience	–	28.1#39	–	Artificial Analysis
MMLU-Pro	–	88.7%#17	–	Vals AI
LegalBench	–	85.0%#27	–	Vals AI
CorpFin	–	71.3%#4	–	Vals AI
TaxEval	–	79.7%#2	–	Vals AI
PRBench Financeeffort not stated	–	–	55.0%#3	Scale AI SEAL
PRBench Legaleffort not stated	–	–	57.0%#4	Scale AI SEAL
MultiNRCeffort not stated	–	–	65.6%#1	Scale AI SEAL
Excel Modeling Benchmarknot in index	–	56.4%#44	–	Vals AI
MedScribenot in index	–	88.9%#10	–	Vals AI
Public Benefits Benchnot in index	–	61.3%#29	–	Vals AI
GDP.pdfnot in index	15.0%#40	–	15.0%#40	Epoch AI Benchmarking Hub
AA-Omniscience: accuracynot in index	–	52.0%#53	–	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	–	50.0%#115	–	Artificial Analysis

Instruction following

Benchmark	xhigh	not stated	Source
LiveBench Languagenot in index	74.3%#52	–	LiveBench
LiveBench Instruction Followingnot in index	69.6%#31	–	LiveBench
MultiChallengeeffort not stated	–	75.3%#3	Scale AI SEAL
Arena Instruction Followingnot in indexeffort not stated	–	1473#31	Arena (LMArena)

Human preference

Benchmark	not stated	Source
Arena Text	1490#13	Arena (LMArena)
EQ-Bench 4	1260#8	EQ-Bench
Arena Creative Writingnot in index	1444#53	Arena (LMArena)
Arena Multi-turnnot in index	1495#15	Arena (LMArena)
Arena Longer Queriesnot in index	1481#45	Arena (LMArena)

Multimodal

Benchmark	xhigh	not stated	Source
Arena Visioneffort not stated	–	1293#26	Arena (LMArena)
MMMU-Pro (Vals)not in index	86.6%#18	–	Vals AI
MortgageTaxnot in index	66.1%#36	–	Vals AI
SAGEnot in index	45.6%#39	–	Vals AI
VTBnot in indexeffort not stated	–	44.8%#1	Scale AI SEAL

Long context

Benchmark	xhigh	not stated	Source	Trend
AA-LCR	77.7%#127	–	Artificial Analysis
Arena Documentnot in indexeffort not stated	–	1480#12	Arena (LMArena)

Composite

Benchmark	xhigh	not stated	Source
Epoch Capabilities Indexnot in indexeffort not stated	–	154.1#43	Epoch AI Benchmarking Hub
LiveBench 1.0not in index	75.3%#37	–	LiveBench
AA Intelligence Index v4.3.2	33.7#101	–	Artificial Analysis
Vals Multimodal Indexnot in index	66.7%#11	–	Vals AI

Safety & honesty

Benchmark	not stated	Source	Trend
FORTRESSnot in index	12.4%#63	Scale AI SEAL

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.

Where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

Provider	Speed	First token	Input $/M	Output $/M	Context	Quantisation
Meta	214 tok/s	1.68 s	$1.25	$4.25	1.0M	–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.150 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0018	$0.0014	2.9 s
Summarize a 30-page report	12,000 / 600	$0.018	$0.0077	4.2 s
Code edit	6,000 / 1,500	$0.014	$0.0089	8.0 s
Agentic coding session	60,000 / 4,000	$0.092	$0.043	18.6 s
Structured extraction	2,000 / 200	$0.0034	$0.0017	2.5 s