Head to head on 18 boards: Google places best on 15, DeepSeek places best on 3 and StepFun places best on 0. All 3 ranked on 5 of 28 boards.
Two at a time on this screen. Tap one to swap it in.
| Board | Read |  StepFun |  Google |  DeepSeek |
|---|
| Chat2 | | | | |
|---|
 Text ArenaArena ratingtodaytoday | Nov 9todaytoday | #701,448Step 5 Preview | #11,526best of the pickedGemini 4 Carbon | #321,475DeepSeek V4.1 Flash |
 MultiChallengeScoretodaytoday | Nov 9todaytoday | | #278.7%best of the pickedGemini 4 Argon | #574.9%DeepSeek V4P1 Flash |
| Reasoning7 | | | | |
|---|
 AA IntelligenceIndextodaytoday | Nov 9todaytoday | #2328.5Step 5 Preview | #160.0best of the pickedGemini 4 Carbon | #1632.6DeepSeek V4.1 Flash |
 GPQA DiamondAccuracytodaytoday | Nov 9todaytoday | | #196.4%best of the pickedGemini 4 Argon | #2193.2%DeepSeek V4.1 Flash |
 MMLU-ProAccuracy · staletodaytoday | Nov 9todaytoday | | #191.2%best of the pickedGemini 3.1 Pro Preview | #3385.0%DeepSeek-V3.2 |
 ARC-AGI-2Scoretodaytoday | Nov 9todaytoday | | #591.3%best of the pickedGemini 4 Argon | #1978.1%DeepSeek V4.1 Flash |
| Epoch ECIECItodaytoday | Nov 9todaytoday | | #6166.4best of the pickedGemini 4 Argon | #19159.3DeepSeek V4.1 Flash |
 LiveBenchGlobal averagetodaytoday | Nov 9todaytoday | | #1578.8%Gemini 3.7 Flash | #881.1%best of the pickedDeepSeek V4.1 Flash |
| HLEAccuracytodaytoday | Nov 9todaytoday | | #346.4%Gemini 3.1 Pro Preview | |
| Coding4 | | | | |
|---|
| WebDev ArenaArena ratingtodaytoday | Nov 9todaytoday | #381,572Step 5 Preview | #41,767best of the pickedGemini 4 Carbon | #141,665DeepSeek V4.1 Flash |
 SWE-bench VerifiedResolved · staletodaytoday | Nov 9todaytoday | | #275.8%best of the pickedGemini 3 Flash | #1170.0%DeepSeek-V3.2 |
 Aider PolyglotPass rate · staletodaytoday | Nov 9todaytoday | | #383.1%best of the pickedGemini 2.5 Pro (Jun 2025) | #874.2%DeepSeek-V3.2 |
| SWE-Bench ProResolvedtodaytoday | Nov 9todaytoday | | #392.2%Gemini 4 Argon | |
| Agents and tools3 | | | | |
|---|
 BFCLOverall accuracy · staletodaytoday | Nov 9todaytoday | | #372.5%best of the pickedGemini 3 Pro | #1256.7%DeepSeek-V3.2 |
| MCP AtlasPass ratetodaytoday | Nov 9todaytoday | | #190.7%Gemini 4 Argon | |
| Search ArenaArena ratingtodaytoday | Nov 9todaytoday | | #141,208Gemini 3.1 Pro Preview | |
| Vision2 | | | | |
|---|
| Vision ArenaArena ratingtodaytoday | Nov 9todaytoday | #511,247Step 5 Preview | #61,294best of the pickedGemini 3.7 Flash | #341,263DeepSeek V4.1 Flash |
| Document ArenaArena ratingtodaytoday | Nov 9todaytoday | | #161,469Gemini 3.8 Flash | |
| Image2 | | | | |
|---|
| Image EditArena ratingtodaytoday | Nov 9todaytoday | #51996Step1x Edit | #81,427best of the pickedGemini Nano Banana 2.1 | |
| Text-to-ImageArena ratingtodaytoday | Nov 9todaytoday | | #71,325Gemini Nano Banana 2.1 | |
| Video2 | | | | |
|---|
| Text-to-VideoArena ratingtodaytoday | Nov 9todaytoday | | #11,541Gemini Omni 1.2 Flash | |
| Image-to-VideoArena ratingtodaytoday | Nov 9todaytoday | | #11,502Gemini Omni 1.2 Flash | |
| Speech2 | | | | |
|---|
| Speech ArenaArena ratingtodaytoday | Nov 9todaytoday | #181,216Stepaudio 2.5 TTS | #61,283best of the pickedGemini 3.8 Flash TTS | |
 Open ASRAverage WERtodaytoday | Nov 9todaytoday | | #5813.20 WERgemma-4-E4B-it (google) | |
| Embeddings1 | | | | |
|---|
| MTEBMean task scoretodaytoday | Nov 9todaytoday | | #568.4%gemini-embedding-001 (google) | |
| Speed1 | | | | |
|---|
| AA SpeedOutput speedtodaytoday | Nov 9todaytoday | #2214 tok/sStep 5 Preview | #735 tok/sbest of the pickedGemini 4 Carbon | #1323 tok/sDeepSeek V4 Pro 0813 |
| Price1 | | | | |
|---|
| AA PriceBlended pricetodaytoday | Nov 9todaytoday | | #5$1.50Gemini 3.8 Flash | #3$0.53best of the pickedDeepSeek V4.1 Flash |
| Usage1 | | | | |
|---|
 OpenRouterWeekly tokenstodaytoday | Nov 9todaytoday | #57.2T tokStep 5 Preview | | #156.0T tokbest of the pickedDeepSeek V4.1 Flash |
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.
Modelsby Glyph