Skip to content

Anthropic vs DeepSeek vs Meta vs MiniMax

Each lab shows its best-ranked model on every board. One row per board where at least one is ranked, from each board's current capture.

  • Anthropic
  • DeepSeek
  • Meta
  • MiniMax

Head to head on 21 boards: Anthropic places best on 13, DeepSeek places best on 3, Meta places best on 3 and MiniMax places best on 2. All 4 ranked on 9 of 27 boards.

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

BoardAnthropicDeepSeekMetaMiniMax
Chat2
Text ArenaArena ratingtoday#11,509best of the pickedClaude Opus 5.5#251,477DeepSeek V4.1 Flash#61,496Muse Spark 1.2#761,440MiniMax-M3
MultiChallengeScoretoday#1259.0%Claude Opus 4.5#2446.1%DeepSeek V3P1#175.5%best of the pickedMuse Spark
Reasoning7
AA IntelligenceIndextoday#157.6best of the pickedClaude Opus 5.5#1839.5DeepSeek V4.1 Flash#448.1Muse Spark 1.3#2829.2MiniMax-M3
LiveBenchGlobal averagetoday#183.4%best of the pickedClaude Fable 5.1#681.1%DeepSeek V4.1 Flash#581.6%Muse Spark 1.3#5567.3%MiniMax-M3
Epoch ECIECItoday#2165.0best of the pickedClaude Fable 5.1#24155.4DeepSeek V4 Pro 0813#15156.9Muse Spark 1.3#60147.0MiniMax-M3
ARC-AGI-2Scoretoday#293.3%best of the pickedClaude Opus 5.5#2361.4%DeepSeek V4 Flash Vision#710.0%Llama 4 Maverick Instruct#524.9%MiniMax-M2.5
MMLU-ProAccuracy · staletoday#489.1%best of the pickedClaude Opus 4.6#3385.0%DeepSeek-V3.2#4782.8%Llama4 Behemoth#688.0%MiniMax M2.1
GPQA DiamondAccuracytoday#1093.9%best of the pickedClaude Opus 5#2091.7%DeepSeek V4 Pro 0813#3589.8%Muse Spark#2590.9%MiniMax-M3
HLEAccuracytoday#246.5%best of the pickedClaude Fable 5.1#640.6%Muse Spark
Coding4
WebDev ArenaArena ratingtoday#11,827best of the pickedClaude Opus 5.5#131,621DeepSeek V4.1 Flash#81,656Muse Spark 1.3#461,483MiniMax-M3
SWE-bench VerifiedResolved · staletoday#176.8%best of the pickedClaude Opus 4.5#1170.0%DeepSeek-V3.2#3621.0%Llama 4 Maverick Instruct#275.8%MiniMax-M2.5
Aider PolyglotPass rate · staletoday#1072.0%Claude Opus 4#874.2%best of the pickedDeepSeek-V3.2#4315.6%Llama 4 Maverick Instruct
SWE-Bench ProResolvedtoday#198.0%Claude Opus 5
Agents and tools3
BFCLOverall accuracytoday#177.5%best of the pickedClaude Opus 4.5#1256.7%DeepSeek-V3.2#3637.3%Llama 4 Maverick 17b 128e
MCP AtlasPass ratetoday#287.2%Claude Fable 5.1#188.1%best of the pickedMuse Spark 1.1
Search ArenaArena ratingtoday#21,253Claude Opus 4.6
Vision2
Vision ArenaArena ratingtoday#11,310best of the pickedClaude Fable 5#51,294Muse Spark 1.3#471,237MiniMax-M3
Document ArenaArena ratingtoday#11,516best of the pickedClaude Opus 5#121,471Muse Spark 1.3#271,435MiniMax-M3
Image2
Image EditArena ratingtoday#61,402Muse Image
Text-to-ImageArena ratingtoday#71,276Muse Image
Video2
Text-to-VideoArena ratingtoday#91,456Muse Video#81,460best of the pickedMiniMax H3
Image-to-VideoArena ratingtoday#11,495MiniMax H3
Speech2
Speech ArenaArena ratingtoday#89847Metavoice V1#161,172best of the pickedSpeech 2.8 Hd
Open ASRAverage WERtoday#284.84 WERmuse-voice-transcribe (meta)
Speed1
AA SpeedOutput speedtoday#988 tok/sClaude Opus 5.5#3232 tok/sDeepSeek V4.1 Flash#2232 tok/sbest of the pickedMuse Spark 1.3
Price1
AA PriceBlended pricetoday#17$4.00Claude Sonnet 5#4$0.53best of the pickedDeepSeek V4.1 Flash#11$2.00Muse Spark 1.3
Usage1
OpenRouterWeekly tokenstoday#191.5T tokClaude Sonnet 5#118.6T tokbest of the pickedDeepSeek V4.1 Flash#151.8T tokMuse Spark 1.3 Contributor

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.

More comparisons

New #1s by email

Saturdays, only in weeks when a leaderboard has a new #1.

Double opt-in. Unsubscribe any time.