Every AI leaderboard

28 leaderboards · 986 models ranked

Alibaba vs Microsoft vs Google vs OpenAI

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.

  • Alibaba
  • Microsoft
  • Google
  • OpenAI

Head to head on 27 boards: Google places best on 11, OpenAI places best on 11, Alibaba places best on 3 and Microsoft places best on 2. All 4 ranked on 6 of 28 boards.

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

BoardAlibabaMicrosoftGoogleOpenAI
Chat2
Text ArenaArena ratingtoday#201,483Qwen3.8 Max (0902)#11,526best of the pickedGemini 4 Carbon#191,485GPT-5.6 Sol
MultiChallengeScoretoday#3341.2%Qwen3-235B-A22B-Thinking (Jul 2025)#278.7%Gemini 4 Argon#182.0%best of the pickedGPT-6.1 Sol
Reasoning7
MMLU-ProAccuracy · staletoday#787.8%Qwen3.5 397B A17B#7576.0%Phi 4 Plus#191.2%best of the pickedGemini 3.1 Pro Preview#389.3%o1
AA IntelligenceIndextoday#1533.7Qwen3.8 Max (0902)#160.0best of the pickedGemini 4 Carbon#356.9GPT-6 Astra
GPQA DiamondAccuracytoday#2692.7%Qwen3.8 Max (0902)#196.4%best of the pickedGemini 4 Argon#295.8%GPT-6 Astra
Epoch ECIECItoday#27157.4Qwen3.8 Max (0902)#6166.4Gemini 4 Argon#1169.1best of the pickedGPT-6 Astra
ARC-AGI-2Scoretoday#4342.4%Qwen3.8 27B#591.3%Gemini 4 Argon#195.0%best of the pickedGPT-6 Astra
LiveBenchGlobal averagetoday#1678.5%Qwen3.8 Max (0902)#1578.8%Gemini 3.7 Flash#482.2%best of the pickedGPT-6 Astra
HLEAccuracytoday#346.4%Gemini 3.1 Pro Preview#154.8%best of the pickedGPT-6 Astra
Coding4
Aider PolyglotPass rate · staletoday#1859.6%Qwen3 235b A22B Diff No Alibaba#383.1%Gemini 2.5 Pro (Jun 2025)#188.0%best of the pickedGPT-5
SWE-bench VerifiedResolved · staletoday#2355.4%Qwen3-Coder 480B/A35B Instruct#275.8%best of the pickedGemini 3 Flash#672.8%GPT-5.2
WebDev ArenaArena ratingtoday#121,677Qwen3.8 Max (0902)#41,767Gemini 4 Carbon#31,770best of the pickedGPT-6 Astra
SWE-Bench ProResolvedtoday#392.2%best of the pickedGemini 4 Argon#490.2%GPT-6 Astra
Agents and tools3
BFCLOverall accuracy · staletoday#2052.1%Qwen3-235B-A22B-Thinking (Jul 2025)#5028.8%Phi 4#372.5%best of the pickedGemini 3 Pro#763.0%o3
MCP AtlasPass ratetoday#784.5%Qwen3.8 2.4T A95B#190.7%best of the pickedGemini 4 Argon#1481.8%GPT-5.6
Search ArenaArena ratingtoday#61,228Qwen3.8 Max (0902)#141,208Gemini 3.1 Pro Preview#11,255best of the pickedGPT-5.6 Sol
Vision2
Vision ArenaArena ratingtoday#21,300best of the pickedQwen3.8 Max (0902)#61,294Gemini 3.7 Flash#81,293GPT-6.1 Sol
Document ArenaArena ratingtoday#311,446Qwen3.7 Plus#161,469Gemini 3.8 Flash#51,486best of the pickedGPT-5.6 Sol
Image2
Text-to-ImageArena ratingtoday#131,254Qwen Image 3.0 Pro#61,330MAI Image 2.6#71,325Gemini Nano Banana 2.1#11,429best of the pickedGPT Image 2.5 Sunburst
Image EditArena ratingtoday#201,364Qwen Image 2.1#41,467MAI Image 2.7#81,427Gemini Nano Banana 2.1#11,517best of the pickedGPT Image 2.5 Sunburst
Video2
Text-to-VideoArena ratingtoday#81,477Wan 3.0#11,541best of the pickedGemini Omni 1.2 Flash#131,367Sora 2 Pro
Image-to-VideoArena ratingtoday#51,481Wan 3.0#11,502best of the pickedGemini Omni 1.2 Flash
Speech2
Speech ArenaArena ratingtoday#31,303best of the pickedQwen Audio 3.1 TTS Plus#311,145Azure Hd 2.5#61,283Gemini 3.8 Flash TTS#421,107TTS 1 Hd
Open ASRAverage WERtoday#106.15 WERQwen3-ASR-1.7B-hf (Qwen)#35.55 WERbest of the pickedazure-speech-09-2026 (microsoft)#5813.20 WERgemma-4-E4B-it (google)#186.80 WERgpt-transcribe (openai)
Embeddings1
MTEBMean task scoretoday#470.6%Qwen3-Embedding-8B (Qwen)#174.3%best of the pickedharrier-oss-v1-27b (microsoft)#568.4%gemini-embedding-001 (google)
Speed1
AA SpeedOutput speedtoday#260 tok/sbest of the pickedQwen3.8 Max (0902)#735 tok/sGemini 4 Carbon#1522 tok/sGPT-6.1 Sol
Price1
AA PriceBlended pricetoday#8$3.00Qwen3.8 Max (0902)#5$1.50best of the pickedGemini 3.8 Flash#6$2.00GPT-6.1 Terra
Usage1
OpenRouterWeekly tokenstoday#47.2T tokGPT-6 Luna

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