Claude Fable 5.1 vs GPT-6 Astra vs Kimi K3 vs GLM-5.3

Head to head on 14 boards: Claude Fable 5.1 places best on 7, GPT-6 Astra places best on 5, GLM-5.3 places best on 2 and Kimi K3 places best on 0. All 4 ranked on 8 of 15 boards.

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

Claude Fable 5.1GPT-6 AstraKimi K3GLM-5.3

Board	Read	Claude Fable 5.1	GPT-6 Astra	Kimi K3	GLM-5.3
Chat1
Text ArenaArena ratingtodaytoday	Sep 27todaytoday	#51,501best of the picked	#231,478	#131,488	#211,480
Reasoning6
Epoch ECIECItodaytoday	Sep 27todaytoday	#2165.0	#1166.6best of the picked	#12157.7	#22155.6
LiveBenchGlobal averagetodaytoday	Sep 27todaytoday	#183.4%best of the picked	#482.2%	#1279.2%	#2876.1%
AA IntelligenceIndextodaytoday	Sep 27todaytoday	#253.4best of the picked	#352.7	#1243.6	#944.8
GPQA DiamondAccuracytodaytoday	Sep 27todaytoday		#195.8%best of the picked	#1593.1%	#2590.9%
ARC-AGI-2Scoretodaytoday	Sep 27todaytoday	#590.0%	#195.0%best of the picked	#2660.4%
HLEAccuracytodaytoday	Sep 27todaytoday	#246.5%	#154.8%best of the picked
Coding2
WebDev ArenaArena ratingtodaytoday	Sep 27todaytoday	#31,751	#21,792best of the picked	#71,660	#151,619
SWE-Bench ProResolvedtodaytoday	Sep 27todaytoday	#292.2%best of the picked	#390.2%	#488.2%	#784.3%
Agents and tools1
MCP AtlasPass ratetodaytoday	Sep 27todaytoday	#287.2%best of the picked		#882.3%	#584.2%
Vision2
Document ArenaArena ratingtodaytoday	Sep 27todaytoday	#21,513best of the picked	#131,468
Vision ArenaArena ratingtodaytoday	Sep 27todaytoday	#91,289best of the picked	#141,284
Speed1
AA SpeedOutput speedtodaytoday	Sep 27todaytoday	#1569 tok/s	#1763 tok/s	#2438 tok/s	#1186 tok/sbest of the picked
Price1
AA PriceBlended pricetodaytoday	Sep 27todaytoday	#23$20.0	#23$20.0	#21$6.00	#12$2.15best of the picked
Usage1
OpenRouterWeekly tokenstodaytoday	Sep 27todaytoday				#102.9T tok

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.