Skip to content

Meta · Chat

Muse Spark 1.3

Meta 🇺🇸ChatMuse familyOne of its sources says the weights aren't publicly available.
From Empiriolabs's listing:
Meta's long-horizon agentic reasoning model with a 1,048,576-token context, image, video, and PDF understanding, web search, and tool calling.
llminfo IQ
140.3
Maker's price
$2.00 blended / 1M
Max context
1M
Listings
26

Evidence and sources

llminfo IQ 140.3, ranked among chat models. From 224 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 125.1, 95% interval 119–132. From 15 benchmark families across 7 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedFair value (52.1). Costs per task about what its benchmark results predict. From 29 benchmarks with measured task costs. Plenty of data behind it.How Value for money worksYap score: tokens used per coding task vs. models with the same resultsChatty (59 ±9). Generates more tokens than models with similar coding results. Measured at max effort. Little data behind it, so treat it as a rough estimate. By effort: default 49 · max 59.Median 265,112 output tokens per task at max effort.How the Yap score worksReleased 2 Sep 2026Maker HQ: US (according to Meta's own website)
All rankings: #99 on OpenRouter this month #100 on OpenRouter this week #140 on OpenRouter today #275 on NanoGPT recentlySources: LMArena data, EmpirioLabs data, Models.dev data, llminfo researchPrimary data from LMArena from EmpirioLabs: primary mode, description from Models.dev: weights availability from llminfo research: release date

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Trade-offs

No open weights

The weights aren't public, so you can't run it yourself.

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 04:34 UTC.

Input & output

as stated by the model's providers
Text
Images
Audio
Video
Files
Text

Capabilities

Supported operations and provider-stated features

VisionReasoningFunction callingStructured outputPrompt cachingPDF inputWeb search

Listings

 Metameta/muse-spark-1.3Sources: LiteLLM data, Models.dev data1M$1.25cache read $0.15$4.25——
 Kenarimuse-spark-1-3Source: Models.dev data1M$1.14cache read $0.1143$4.57——
 GoogleRegions: globalvertex_ai/meta/muse-spark-1.3Source: LiteLLM data1M$1.25cache read $0.15$4.25——
 Eden AIInference by Metameta/muse-spark-1.3Source: Models.dev data1M$1.25cache read $0.15$4.25——
 Kilometa/muse-spark-1.3Sources: Kilo AI Gateway data, Models.dev data1M$1.25cache read $0.15$4.25——
 Llmgatewaymuse-spark-1.3Source: LLMGateway data1M$1.25cache read $0.15$4.25——
 NanoGPTmeta/muse-spark-1.3Sources: Models.dev data, NanoGPT1M$1.25cache read $0.15$4.25——
 Nous Portalmeta/muse-spark-1.3Source: Nous Portal Inference Gateway data1M$1.25cache read $0.15$4.25——
 OpenRouterInference by Googlemeta/muse-spark-1.3 · globalSource: OpenRouter data docs1M$1.25cache read $0.15$4.251751,533ms
 Vercelmeta/muse-spark-1.3Sources: Models.dev data, Vercel AI Gateway data1M$1.25cache read $0.15$4.25——
Show all 23 listings

Showing 10 of 23: the maker's own listing first, then the cheapest per 1M input tokens from each provider.

Also sold by Aihubmix, Aimlapi, Cloudflare AI Gateway, Commandcode, Empiriolabs, Llmgateway Providers, Neosantara, Opper, Wavespeed, Zenmux.

3 listings without a price are hidden.

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Human preference

Arena ELO59 related benchmarks · 67 results
Source: LMArena data · Nov 2026
BenchmarkxHighMax
Arena ELONot measured
ChineseNot measured
Creative WritingNot measured
DocumentNot measured
Show 55 more benchmarksHide 55 benchmarks
BenchmarkxHighMax
Document Style-ControlledNot measured
EnglishNot measured
Exclude TiesNot measured
ExpertNot measured
FrenchNot measured
GermanNot measured
Hard PromptsNot measured
Hard Prompts EnglishNot measured
Industry Business and Management and Financial OperationsNot measured
Industry Entertainment and Sports and MediaNot measured
Industry Life and Physical and Social ScienceNot measured
Industry MathematicalNot measured
Industry Medicine and HealthcareNot measured
Industry Software and It ServicesNot measured
Industry Writing and Literature and LanguageNot measured
JapaneseNot measured
KoreanNot measured
Longer QueryNot measured
Multi TurnNot measured
Non EnglishNot measured
PolishNot measured
RussianNot measured
SpanishNot measured
Style-ControlledNot measured
Text FactualityNot measured
Text Factuality - ChineseNot measured
Text Factuality - Creative WritingNot measured
Text Factuality - EnglishNot measured
Text Factuality - Exclude TiesNot measured
Text Factuality - ExpertNot measured
Text Factuality - FrenchNot measured
Text Factuality - Hard PromptsNot measured
Text Factuality - Hard Prompts EnglishNot measured
Text Factuality - Industry Business and Management and Financial OperationsNot measured
Text Factuality - Industry Entertainment and Sports and MediaNot measured
Text Factuality - Industry Life and Physical and Social ScienceNot measured
Text Factuality - Industry MathematicalNot measured
Text Factuality - Industry Medicine and HealthcareNot measured
Text Factuality - Industry Software and It ServicesNot measured
Text Factuality - Industry Writing and Literature and LanguageNot measured
Text Factuality - Longer QueryNot measured
Text Factuality - Multi TurnNot measured
Text Factuality - Non EnglishNot measured
Text Factuality - RussianNot measured
Text Factuality - SpanishNot measured
Vision
Vision - Chinese
Vision - Creative Writing Vision
Vision - Diagram
Vision - English
Vision - Homework
Vision - Humor
Vision Style-Controlled
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026

Agentic

Agentic IF Indexmax · Meta internal composite · agentic instruction following · Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Arena AgentSource: LMArena data
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026
GDPval-AASource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
GDPval-AA v2max · Artificial Analysis Stirrup · 220 tasks · shell + web browsing · human baseline 1000 · Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
JobBench2 results
Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · 65 tasks · official evaluation code · OpenCode · file-aware rubric grader · mean rubric score
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
OSWorld4 related benchmarks · 4 results
2.0 · Binary
Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · OSWorld 2.0 v08.08 · 108 tasks · common internal framework · binary completion
2.0 · Partial
Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · OSWorld 2.0 v08.08 · 108 tasks · common internal framework · partial reward
OSWorld 2.0
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
SWE-Atlas · Codebase Q&Amax · public QnA split · 124 tasks · mini-swe-agent · rubric grading · mean pass@1 · Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · 89 tasks · native coding harness · isolated cloud sandbox · pass@1
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Coding

CursorBench 4.0Reasoning effort · Cursor production agent · Reported by Cursor · via BenchBoard docs
ValsSource: Vals AI

General

Reasoning

Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · public v3 · 600 tasks · benchmark automation tools · pass@1
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
DeepSearchQA2 results
Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
max · 900 questions · common browser search/open/find harness · semantic answer matching
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs · Sep 2026

Math

Long context

OpenAI MRCR v2 · 8-needlemax · 100 examples · 8-needle · no tools · mean sequence-matcher ratio · Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Instruction following

OCR

Science

Vision

Other

Artificial AnalysisSource: OpenRouter data docs
DeepSWE v1.1max · mini-swe-agent · 113 tasks · task pass rate · Reported by Meta · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
MentalHealthBench17 related benchmarks · 17 results
Overall
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Acuity · Emergent
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Acuity · High acuity
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Acuity · Non-acute
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Actionable guidance and coping
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Alarmism and under-response
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Show 11 more resultsHide 11 results
Behavior · Collaboration and user agency
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Communication and writing
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Context seeking and assessment
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Empathy and emotional support
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Interpretation and cognitive reframing
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Reality-testing calibration
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
Behavior · Refusals and harm avoidance
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
User profile · Caregiver
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
User profile · Clinician
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
User profile · Layperson
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
User profile · Teen
Reported by OpenAI · via BenchBoard docs
OpenAI evaluation · 95% CI available on source page
ValsSource: Vals AI

Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · LMArena, CC BY 4.0 (licence) · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence) · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · MineBench, MIT (licence) · Steel.dev Agent Leaderboards, MIT (licence) · UzLiB, Tahrirchi (Abror Shopulatov). Public LEADERBOARD.md numbers; repository MIT covers software and documentation, no separate results licence stated and none asserted. · WeirdML v3 by Håvard Tveit Ihle (Norwegian Defence Research Establishment). API costs supported by Epoch AI, METR and NDRE.

Around the web

Also known as: empiriolabs/muse-spark-1-3, meta/muse-spark-1.3, muse-spark-1-3, muse-spark-1.3-max, muse-spark-1.3-xhigh, vertex_ai/meta/muse-spark-1.3

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works