Claude Sonnet 5.5 vs Muse Spark vs Kimi K3 vs Gemini 3.8 Flash

Head to head on 14 boards: Gemini 3.8 Flash places best on 6, Claude Sonnet 5.5 places best on 4, Kimi K3 places best on 3 and Muse Spark places best on 1. All 4 ranked on 3 of 16 boards.

Two at a time on this screen. Tap one to swap it in.

Claude Sonnet 5.5Muse SparkKimi K3Gemini 3.8 Flash

Board	Read	Claude Sonnet 5.5	Muse Spark	Kimi K3	Gemini 3.8 Flash
Chat2
Text ArenaArena ratingtodaytoday	Oct 5todaytoday	#391,471	#121,489	#131,488	#71,495best of the picked
MultiChallengeScoretodaytoday	Oct 5todaytoday		#175.5%
Reasoning6
GPQA DiamondAccuracytodaytoday	Oct 5todaytoday	#295.6%best of the picked	#4189.8%	#1893.1%	#395.4%
Epoch ECIECItodaytoday	Oct 5todaytoday	#3165.2best of the picked	#40152.1	#13157.6	#15156.9
AA IntelligenceIndextodaytoday	Oct 5todaytoday	#256.0best of the picked		#1343.6	#1640.9
LiveBenchGlobal averagetodaytoday	Oct 5todaytoday	#2077.8%		#1379.2%best of the picked	#3275.8%
HLEAccuracytodaytoday	Oct 5todaytoday		#640.6%		#444.5%best of the picked
ARC-AGI-2Scoretodaytoday	Oct 5todaytoday			#2960.4%	#889.2%best of the picked
Coding2
WebDev ArenaArena ratingtodaytoday	Oct 5todaytoday	#31,786best of the picked		#101,658	#251,583
SWE-Bench ProResolvedtodaytoday	Oct 5todaytoday			#488.2%best of the picked	#958.8%
Agents and tools1
MCP AtlasPass ratetodaytoday	Oct 5todaytoday		#982.2%	#882.3%best of the picked
Vision2
Vision ArenaArena ratingtodaytoday	Oct 5todaytoday	#291,268	#61,294best of the picked		#91,290
Document ArenaArena ratingtodaytoday	Oct 5todaytoday		#241,444
Speed1
AA SpeedOutput speedtodaytoday	Oct 5todaytoday	#5139 tok/s		#2536 tok/s	#1249 tok/sbest of the picked
Price1
AA PriceBlended pricetodaytoday	Oct 5todaytoday	#17$4.00		#22$6.00	#10$1.50best of the picked
Usage1
OpenRouterWeekly tokenstodaytoday	Oct 5todaytoday			#151.6T tok	#132.2T tokbest of the picked

Green marks the best placing among the picked where two or more are ranked; ties are marked for each. Ranks are each board's own order (lower is better), scores as the board published them. Dash: not on that board.