Head-to-head comparison

Muse Wonder vs. Kimi K2.6

Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.

shared result series: 131
distinct benchmarks: 10
latest retrieval: 2026-10-15

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for Muse Wonder and Kimi K2.6
Attribute	Muse Wonder	Kimi K2.6
Provider	Meta	Moonshot AI
Released	Oct 12, 2026	Apr 20, 2026
Availability	Preview	Active
Model type	Proprietary	Open Weights
Parameters	Not published	1T, 32B active
Architecture	Not published	Mixture of Experts
Context window	1,048,576 tokens	262,144 tokens
Knowledge cutoff	Not published	Not published
Technical sources	Model, Context	Model, Context

Pricing

API pricing compared

The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.

API pricing for Muse Wonder and Kimi K2.6
Price type	Muse Wonder	Kimi K2.6
API input	$4	$0.95
API output	$20	$4
Cache read	$0.2	$0.16
Cache write (5 min.)	Not listed	Not listed
Cache write (1 hr.)	Not listed	Not listed
Price verified	10/15/2026 Source	09/22/2026 Source

Muse WonderKimi K2.6

API inputUSD per 1M tokens, base tier

Muse Wonder$4

Kimi K2.6$0.95

API outputUSD per 1M tokens, base tier

Muse Wonder$20

Kimi K2.6$4

Cache readUSD per 1M tokens, base tier

Muse Wonder$0.2

Kimi K2.6$0.16

Key to the marks:better valuebetter, but not conclusivetie

Pricing by provider

Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.

No provider tariff for Muse Wonder is available in the stored OpenRouter snapshot. The official direct price is still shown.

Provider pricing for Muse Wonder and Kimi K2.6
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
Muse Wonder	Meta	Direct from developerStandard API	Lowest listed price: $4	Lowest listed price: $20	Source10/15/2026
Kimi K2.6	Moonshot AI	Direct from developerStandard API	$0.95	$4	Source09/22/2026
	AtlasCloud	Via OpenRouteratlas-cloud/int4	$0.95	$4	Source10/16/2026
	Baidu	Via OpenRouterbaidu/fp4	$0.95	$4	Source10/16/2026
	Chutes	Via OpenRouterchutes/int4	$0.45	Lowest listed price: $1.95	Source10/16/2026
	Cloudflare	Via OpenRoutercloudflare	$0.95	$4	Source10/16/2026
	CoreWeave	Via OpenRoutercoreweave/fp4	$0.65	$3.41	Source10/16/2026
	Crusoe	Via OpenRoutercrusoe/bf16	$0.7	$3.5	Source10/16/2026
	DeepInfra	Via OpenRouterdeepinfra/fp4	$0.75	$3.5	Source10/16/2026
	DigitalOcean	Via OpenRouterdigitalocean	$0.57	$2.4	Source10/16/2026
	GMICloud	Via OpenRoutergmicloud/fp8	$0.855	$3.6	Source10/16/2026
	Inceptron	Via OpenRouterinceptron/int4	Lowest listed price: $0.4323	$3.2	Source10/16/2026
	Moonshot AI	Via OpenRoutermoonshotai/int4	$0.95	$4	Source10/16/2026
	Parasail	Via OpenRouterparasail/int4	$0.75	$3.5	Source10/16/2026
	Phala	Via OpenRouterphala	$1.09	$4.6	Source10/16/2026
	SiliconFlow	Via OpenRoutersiliconflow/fp8	$0.77	$3.4	Source10/16/2026
	StreamLake	Via OpenRouterstreamlake/fp8	$0.5985	$2.52	Source10/16/2026
	Venice	Via OpenRoutervenice/int4	$0.75	$3.5	Source10/16/2026

Performance

Shared benchmarks

Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.

Capability profile from matched benchmarks

Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.

Not enough directly matched categories for a reliable radar

Only 0 of at least 5 categories have comparable, direct model measurements. The benchmark tables below show the available evidence without filling missing axes.

Muse WonderKimi K2.6

Code MigrationOverall. Higher is better. Test setup not documented as identical

Muse Wonder62.552 %

Kimi K2.627.768 %

EMBOverall. Higher is better. Test setup not documented as identical

Muse Wonder67.084 %

Kimi K2.657.855 %

Finance Agent (v2)Overall. Higher is better. Test setup not documented as identical

Muse Wonder56.637 %

Kimi K2.644.899 %

Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. Test setup not documented as identical

Muse Wonder17.083 %

Kimi K2.61.667 %

Legal Research BenchOverall · All-pass. Higher is better. Test setup not documented as identical

Muse Wonder50.481 %

Kimi K2.615.865 %

LMArena Code Arena WebDevOverall. Higher is better. Test setup not documented as identical

Muse Wonder1,650.947

Kimi K2.61,509.237

LMArena Text, Style ControlOverall. Higher is better. Test setup not documented as identical

Muse Wonder1,512.671

Kimi K2.61,460.76

Public Benefits Bench v1.1Overall. Higher is better. Test setup not documented as identical

Muse Wonder70.23 %

Kimi K2.656.631 %

Tax Agent BenchOverall. Higher is better. Test setup not documented as identical

Muse Wonder36.269 %

Kimi K2.612.404 %

Vibe Code Bench v1.1Overall. Higher is better. Test setup not documented as identical

Muse Wonder83.089 %

Kimi K2.637.891 %

Code MigrationCLI. Higher is better. Test setup not documented as identical

Muse Wonder60.07 %

Kimi K2.617.977 %

EMBComps. Higher is better. Test setup not documented as identical

Muse Wonder79.488 %

Kimi K2.661.395 %

Key to the marks:better valuebetter, but not conclusivetie

Benchmark scores for Muse Wonder and Kimi K2.6
Benchmark and task	Muse Wonder	Kimi K2.6	Source
Code MigrationOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value62.552 %	27.768 %	Vals AI Source details Data as of: 10/12/2026
EMBOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value67.084 %	57.855 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Overall Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value56.637 %	44.899 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkOverall · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value17.083 %	1.667 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchOverall · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value50.481 %	15.865 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevOverall Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Code Arena WebDev Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Code Arena WebDevMuse Wonder: Arena rating from 1,228 published pairwise votes.Kimi K2.6: Arena rating from 10,192 published pairwise votes.	Higher measured value1,650.947 95% confidence interval: ±18.263The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,509.237 95% confidence interval: ±7.227The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlOverall Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 1,944 published pairwise votes.Kimi K2.6: Arena rating from 40,030 published pairwise votes.	Higher measured value1,512.671 95% confidence interval: ±13.236The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,460.76 95% confidence interval: ±4.428The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Public Benefits Bench v1.1Overall Test details Test setup not documented as identicalNo direct winnerVersion: 1.1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value70.23 %	56.631 %	Vals AI Source details Data as of: 10/12/2026
Tax Agent BenchOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value36.269 %	12.404 %	Vals AI Source details Data as of: 10/12/2026
Vibe Code Bench v1.1Overall Test details Test setup not documented as identicalNo direct winnerVersion: 1.1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Harness: OpenHands, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Harness: OpenHands, Top P: 0.95, Output limit: 128,000 tokens	Higher measured value83.089 %	37.891 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCLI Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value60.07 %	17.977 %	Vals AI Source details Data as of: 10/12/2026
EMBComps Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value79.488 %	61.395 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Adjustments Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value52.67 %	37.934 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value93.841 %	63.768 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchAdministrative / Regulatory · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value63.636 %	18.182 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevWebdev Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Code Arena WebDev Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Code Arena WebDevMuse Wonder: Arena rating from 1,228 published pairwise votes.Kimi K2.6: Arena rating from 10,192 published pairwise votes.	Higher measured value1,650.947 95% confidence interval: ±18.263The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,509.237 95% confidence interval: ±7.227The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlBusiness, management and finance Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 338 published pairwise votes.Kimi K2.6: Arena rating from 7,937 published pairwise votes.	Higher measured value1,500.247 95% confidence interval: ±30.856The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,454.55 95% confidence interval: ±7.536The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchControversy & Precedence Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value20 %	3.333 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCOBOL Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value70 %	57.143 %	Vals AI Source details Data as of: 10/12/2026
EMBDataroom Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value79.232 %	75.069 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)All-Pass Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value45.936 %	32.584 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value20 %	0 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchAdministrative / Regulatory · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value90.571 %	75.531 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevWebdev-html Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Code Arena WebDev Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Code Arena WebDevMuse Wonder: Arena rating from 158 published pairwise votes.Kimi K2.6: Arena rating from 1,160 published pairwise votes.	Higher measured value1,721.908 95% confidence interval: ±51.47The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,484.407 95% confidence interval: ±18.204The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlCoding Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 487 published pairwise votes.Kimi K2.6: Arena rating from 11,100 published pairwise votes.	Higher measured value1,557.956 95% confidence interval: ±26.189The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,515.795 95% confidence interval: ±6.881The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchControversy & Precedence Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value71.362 %	42.296 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCode Quality Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: mean reviewer code-quality ratingScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value4.061 / 5	3.149 / 5	Vals AI Source details Data as of: 10/12/2026
EMBDCF Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value51.035 %	49.675 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Comparables Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value47.933 %	41.069 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkArbitration International Dispute Resolution · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value92.473 %	50.179 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchBusiness & Commercial · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value48.837 %	20.93 %	Vals AI Source details Data as of: 10/12/2026
LMArena Code Arena WebDevWebdev-react Test details Test setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Code Arena WebDev Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Code Arena WebDevMuse Wonder: Arena rating from 787 published pairwise votes.Kimi K2.6: Arena rating from 7,354 published pairwise votes.	Higher measured value1,637.813 95% confidence interval: ±23.199The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,504.994 95% confidence interval: ±7.997The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlCreative writing Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 427 published pairwise votes.Kimi K2.6: Arena rating from 6,330 published pairwise votes.	Higher measured value1,485.682 95% confidence interval: ±29.474The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,432.668 95% confidence interval: ±8.216The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchCurrent & Temporal Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value50 %	26.467 %	Vals AI Source details Data as of: 10/12/2026
EMBLBO Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value60.69 %	43.117 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Disclosure Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value67.252 %	52.124 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkArbitration International Dispute Resolution · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	No direct winner0 %	No direct winner0 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchBusiness & Commercial · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value86.88 %	78.436 %	Vals AI Source details Data as of: 10/12/2026
LMArena Text, Style ControlEnglish Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 802 published pairwise votes.Kimi K2.6: Arena rating from 18,177 published pairwise votes.	Higher measured value1,523.395 95% confidence interval: ±20.831The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,469.196 95% confidence interval: ±5.636The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchCurrent & Temporal Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value77.655 %	51.239 %	Vals AI Source details Data as of: 10/12/2026
EMBM&A Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value76.852 %	64.205 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Earnings Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value76.768 %	63.824 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkBanking Finance · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value94.424 %	84.758 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchCivil Litigation · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value37.037 %	18.519 %	Vals AI Source details Data as of: 10/12/2026
LMArena Text, Style ControlEntertainment, sports and media Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 536 published pairwise votes.Kimi K2.6: Arena rating from 8,304 published pairwise votes.	Higher measured value1,468.099 95% confidence interval: ±25.703The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,429.904 95% confidence interval: ±7.52The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchFact-Pattern Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value39.535 %	9.302 %	Vals AI Source details Data as of: 10/12/2026
EMBMisc Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value56.936 %	54.599 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Financial Modeling Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value22.515 %	14.201 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkBanking Finance · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value20 %	0 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchCivil Litigation · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value89.018 %	75.8 %	Vals AI Source details Data as of: 10/12/2026
LMArena Text, Style ControlExcluding ties Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 1,390 published pairwise votes.Kimi K2.6: Arena rating from 29,659 published pairwise votes.	Higher measured value1,538.781 95% confidence interval: ±18.91The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,464.547 95% confidence interval: ±5.8The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchFact-Pattern Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value67.563 %	46.747 %	Vals AI Source details Data as of: 10/12/2026
EMBOperating Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value65.354 %	56.923 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)General Qualitative Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value70.779 %	57.295 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkBankruptcy Restructuring · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value89.965 %	54.325 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchConstitutional / Civil Rights · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value39.474 %	21.053 %	Vals AI Source details Data as of: 10/12/2026
LMArena Text, Style ControlExpert prompts Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 212 published pairwise votes.Kimi K2.6: Arena rating from 3,855 published pairwise votes.	Higher measured value1,512.579 95% confidence interval: ±38.177The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,502.724 95% confidence interval: ±9.922The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchForms & Filings Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value23.333 %	3.333 %	Vals AI Source details Data as of: 10/12/2026
EMBScratch Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value69.002 %	59.678 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)General Quantitative Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value77.605 %	65.13 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkBankruptcy Restructuring · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	No direct winner0 %	No direct winner0 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchConstitutional / Civil Rights · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value89.259 %	79.224 %	Vals AI Source details Data as of: 10/12/2026
LMArena Text, Style ControlHard prompts Test details Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterMuse Wonder: Snapshot: muse-wonder-max, Reasoning: max, Harness: Arena Text, style control Kimi K2.6: Snapshot: kimi-k2.6, Harness: Arena Text, style controlMuse Wonder: Arena rating from 1,218 published pairwise votes.Kimi K2.6: Arena rating from 26,084 published pairwise votes.	Higher measured value1,533.453 95% confidence interval: ±16.637The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	1,485.51 95% confidence interval: ±5.181The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
Tax Agent BenchForms & Filings Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterMuse Wonder: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens Kimi K2.6: Top P: 0.95, Output limit: 128,000 tokens	Higher measured value69.85 %	37.63 %	Vals AI Source details Data as of: 10/12/2026
Key to the marks:better valuebetter, but not conclusivetie

Our model tests

Our tests

There is no shared test round for this selection. Results from different rounds are not mixed.

Earlier tests

Pelican on a bicycle

A community classic for free-form SVG drawing.

Muse Wonder

Not tested yet

Kimi K2.6

Output withheld for safety

Reasoning (requested): high

Current attempt

Run date: Oct 5, 2026
API provider: openrouter
Requested API model: moonshotai/kimi-k2.6
API model reported in response: moonshotai/kimi-k2.6
Reasoning setting: high
Pinned endpoint: Moonshot AI | moonshotai/kimi-k2.6-20260420
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Failed
Response time: 124 s

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Earlier attempts (3)

Oct 2, 2026Output withheld for safetyReasoning setting: high
Oct 1, 2026Output withheld for safetyReasoning setting: high
Sep 7, 2026API request failedReasoning setting: high

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

How to read this comparison

Winning one benchmark is not an overall verdict

Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.

The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.

Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.

More matchups

Compare Muse Wonder and Kimi K2.6 with other leading models using the same data and benchmark logic.

View the complete LLM comparison