BenchLeader

Muse Ember vs Qwen3.5-27B

Full profiles: Muse Ember (Meta) and Qwen3.5-27B (Alibaba): every benchmark, price and rank.

Updated · Data from 7 sources

As of 16 Nov 2026, there are not enough shared benchmarks for a verdict, and Muse Ember is 4.1× cheaper at $0.200 vs $0.825 per 1M tokens blended. Choose Muse Ember for coding (60 vs 44), where it also streams 3×+ faster; choose Qwen3.5-27B for fast first answers (2.9 s vs 5.0 s) and reasoning (58 vs 51).

Key facts (as of 16 Nov 2026)

  • Muse Ember scores 52.1 on the BenchLeader Index and Qwen3.5-27B scores 52.0 (BenchLeader, 16 Nov 2026).
  • Muse Ember costs $0.200 per 1M tokens blended (list price, 16 Nov 2026).
  • Qwen3.5-27B costs $0.825 per 1M tokens blended (list price, 16 Nov 2026).
  • Muse Ember streams 3×+ faster than Qwen3.5-27B (Artificial Analysis, 7-day median, 16 Nov 2026).
  • Of the 1 benchmarks both models have, Muse Ember leads 1 (BenchLeader, 16 Nov 2026).
  • On Arena WebDev, Muse Ember is rated 1609 and Qwen3.5-27B 1357 (LMArena, as published by 16 Nov 2026).

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
CodingMuse Ember60 vs 44 category score; Arena WebDev 1609 vs 1357High
ReasoningQwen3.5-27B58 vs 51 category scoreMedium
Chat quality (human preference)–Only Qwen3.5-27B has an Arena rating (58 category score)Not comparable
Lowest costMuse Ember$0.200 vs $0.825 per 1M blended; agentic coding session $0.008 vs $0.028–
Fastest first answerQwen3.5-27B2.9 s (Qwen3.5-27B) vs 5.0 s (Muse Ember (max)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • Not enough shared benchmarks for a verdict.
  • Muse Ember (max) is stronger in coding.
  • Qwen3.5-27B is stronger in reasoning.
  • Muse Ember (max) is 4.1× cheaper ($0.200 vs $0.825 per 1M blended).
  • Muse Ember (max) streams 11.6× faster (437 vs 38 tokens per second).

At a glance

SpecMuse EmberQwen3.5-27B
DeveloperMetaAlibaba
Released9 Nov 202624 Feb 2026
Context window (tokens)1,048,576262,144
Compared atMuse Ember (max)Qwen3.5-27B
Open weightsNoYes
Price in / out / cached input per 1M$0.100 / $0.500 / $0.0040$0.300 / $2.40 / –

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.

CategoryMuse Ember (max)Qwen3.5-27BLeader
Reasoning51.158.3Qwen3.5-27B
Coding60.044.3Muse Ember
Agents & tools52.4–not comparable
Maths–45.6not comparable
Knowledge48.8–not comparable
Human preference–58.2not comparable
Multimodal–56.8not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsKnowledgeMultimodalHumanpreferenceMuse Ember, Reasoning: 51.1Muse Ember, Coding: 60.0Muse Ember, Agents & tools: 52.4Muse Ember, Knowledge: 48.8Qwen3.5-27B, Reasoning: 58.3Qwen3.5-27B, Coding: 44.3Qwen3.5-27B, Maths: 45.6Qwen3.5-27B, Multimodal: 56.8Qwen3.5-27B, Human preference: 58.2

Dashed: that model has no result on the axis it crosses; the line only joins its neighbours.

Not drawn, no result for these models: Instruction following, Long context.

Model, click to add or removeIndex

Where sources disagree

Across the 1 benchmarks both have, Muse Ember leads 1, Qwen3.5-27B leads 0.

Overall indexMuse Ember (max)Qwen3.5-27BLeader
BenchLeader Index52.152.0neck and neck

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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureMuse Ember (max)Qwen3.5-27B
Blended price per 1M tokens$0.200$0.825
Input / output per 1M$0.100 / $0.500$0.300 / $2.40
Cached input per 1M$0.0040–
Price per task: chat reply$0.0002$0.0008
Price per task: summarize a 30-page report$0.0015$0.0050
Price per task: code edit$0.0014$0.0054
Price per task: agentic coding session$0.0080$0.028
Price per task: structured extraction$0.0003$0.0011
Agentic coding session, input mostly cached$0.0037–
Cost per benchmark task (AA Index)$0.136–
Output tokens per task (AA Index)84k–
Of which reasoning52k–
Time per task (AA Index)3 min–
Output speed437 tok/s38 tok/s
Time to first answer5.0 s2.9 s
Time to a complete answer, last measured 13 Nov 202612.9 s–

Each model is compared at its best-ranked reasoning-effort setting. Muse Ember (max) scores 52.1 and takes 5 s to its first answer token; Muse Ember (xhigh) scores 49.1 in 4 s. Qwen3.5-27B scores 52.0 and takes 3 s to its first answer token. 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

66 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.

MetricPublisherMuse Ember (max)Qwen3.5-27B
BenchLeader IndexBenchLeader52.152.0
Agents & tools scoreBenchLeader52.4–
Coding scoreBenchLeader60.044.3
Knowledge scoreBenchLeader48.8–
Reasoning scoreBenchLeader51.158.3
Human preference scoreBenchLeader–58.2
Maths scoreBenchLeader–45.6
Multimodal scoreBenchLeader–56.8
Blended price $/MBenchLeader$0.200$0.825
Output speedBenchLeader437 tok/s38 tok/s
Time to first answerBenchLeader5.0 s2.9 s
Context windowBenchLeader1,048,576 tokens262,144 tokens
Reasoning
SimpleBenchSimpleBench72.7%†–
Arena Hard PromptsArena (LMArena)–1427
GPQA Diamond (AA)Artificial Analysis–84.2%†
Humanity's Last Exam (AA)Artificial Analysis–13.9%†
HLE-Diamond (AA)Artificial Analysis25.1%–
CritPtArtificial Analysis–0.3%†
CritPt v2 hard subset (AA)Artificial Analysis39.0%–
Arena ExpertArena (LMArena)–1441
LMCAEpoch AI Benchmarking Hub–34.0%
DTBenchEpoch AI Benchmarking Hub–82.4%
Coding
WeirdMLWeirdML–39.5%
Arena CodingArena (LMArena)–1452
Arena WebDevArena (LMArena)16091357
Vibe Code Bench v1.1Vals AI83.7%–
SWE Atlas v1.1Scale AI SEAL56.0%†–
SWE Atlas v1.1: Codebase QnAScale AI SEAL61.3%†–
SWE Atlas v1.1: Test WritingScale AI SEAL56.7%†–
SWE Atlas v1.1: RefactoringScale AI SEAL50.0%†–
ALE-BenchEpoch AI Benchmarking Hub–349.4
Terminal-Bench 4.0 (AA)Artificial Analysis40.9%–
Design Arena WebsiteDesign Arena1330†–
FrontierCode 2.0 (AA)Artificial Analysis27.2%–
Agents & tools
Terminal-BenchTerminal-Bench30.6%†–
APEX-AgentsMercor64.3%†–
Arena AgentArena (LMArena)1.7–
Show all 66 benchmarks
MetricPublisherMuse Ember (max)Qwen3.5-27B
Agents & tools
Terminal-Bench HardArtificial Analysis–31.8%†
τ²-Bench Telecom (AA)Artificial Analysis–87.1%†
ITBench SRE (AA)Artificial Analysis–35.5%†
Finance Agent v2.1Vals AI63.8%–
Harvey's Legal Agent Benchmark v1.1Vals AI25.0%–
Legal Research BenchVals AI49.0%–
Tax Agent BenchVals AI38.2%–
Terminal-Bench 4.0 (Vals)Vals AI39.4%–
Arena Agent Task CompletionArena (LMArena)-7.6–
Vending-Bench 2Andon Labs–202.0
AutomationBenchArtificial Analysis68.1%–
GDP.pdf (AA)Artificial Analysis22.8%–
AA-Briefcase v1.1Artificial Analysis1520–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis20.0%–
Maths
AIME 2026MathArena–91.7%
HMMT February 2026MathArena–81.1%
MathArena ApexMathArena–2.1%
Arena MathsArena (LMArena)–1423
Knowledge
AA-Omniscience v1.1Artificial Analysis7.4–
Excel Modeling Benchmark v1.1Vals AI72.2%–
AA-Omniscience v1.1: accuracyArtificial Analysis21.1%–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis82.6%–
Global-MMLU-Lite (AA)Artificial Analysis–83.8%†
Instruction following
IFBenchArtificial Analysis–46.9%†
Arena Instruction FollowingArena (LMArena)–1400
Human preference
Arena TextArena (LMArena)–1409
Arena Creative WritingArena (LMArena)–1358
Arena Multi-turnArena (LMArena)–1414
Arena Longer QueriesArena (LMArena)–1422
Multimodal
Arena VisionArena (LMArena)–1240
MMMU-ProArtificial Analysis–70.0%†
Long context
AA-LCR v1.1Artificial Analysis–64.0%†
MLCRArtificial Analysis18.3%–
Composite
AA Intelligence Index v5.0.1Artificial Analysis36.0–
AA Intelligence Index v4.3.2Artificial Analysis–19.4†
AA Progress Index v1Artificial Analysis–250.2†
Vals IndexVals AI60.5–
Vals Finance IndexVals AI60.9%–
Vals Legal IndexVals AI75.7%–
Safety & honesty
FORTRESSScale AI SEAL11.9%†–
Metadata
AA Openness IndexArtificial Analysis–38.9†

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

What changed

  • Qwen3.5-27B: BenchLeader Index 52.5 → 52.0.
  • Muse Ember (max): 1 new result (Arena Agent).
  • Muse Ember (max): 2 new results (MLCR, Arena WebDev).

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

Muse Ember vs Qwen3.5-27B: questions

Is Muse Ember better than Qwen3.5-27B?
Not enough shared benchmarks for a verdict. The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
Muse Ember vs Qwen3.5-27B: which is cheaper, and will that change?
Muse Ember is cheaper today: $0.200 against $0.825 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Muse Ember or Qwen3.5-27B better for coding?
Muse Ember scores higher in coding (60 vs 44 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1609 for Muse Ember against 1357 for Qwen3.5-27B.
Which is faster, Muse Ember or Qwen3.5-27B?
Muse Ember streams 3×+ faster (437 vs 38 output tokens per second, Artificial Analysis 7-day median).
Which is better for long documents?
Muse Ember accepts more context: 1,048,576 against 262,144 tokens for Qwen3.5-27B.
More questions (3)
Which has lower latency?
Qwen3.5-27B answers first: 2.9 s against 5.0 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on Muse Ember vs Qwen3.5-27B?
Across the 1 benchmarks both have, Muse Ember leads 1, Qwen3.5-27B leads 0.
When was Muse Ember released?
Muse Ember was released on 9 Nov 2026.