LMArena has been renamed Arena, with a UI refresh.

https://arena.ai/blog/lmarena-is-now-arena/

#lmarena

2,4 Billionen Parameter: Baidus Ernie 5.0 schiebt sich im LMArena-Ranking auf Platz 8. Das Modell übertrifft mit einem Score von 1459 knapp OpenAIs GPT-5.1. Technisch setzt Baidu auf native Multimodalität und verarbeitet Bild- und Textinformationen im selben Kontextraum. Trotz US-Sanktionen verkleinert sich der Abstand zu Spitzenreitern wie Googles Gemini-3-pro drastisch. #Baidu #Ernie5 #LMArena
https://www.all-ai.de/news/news26top/ernie-5-top
Ernie 5.0 Release: So mächtig ist Chinas neues Supermodell

Der neue Spitzenreiter für den chinesischen Raum zeigt beeindruckende multimodale Fähigkeiten und definiert KI-Leistung in Asien völlig neu.

All-AI.de

ʟᴇɢɪᴛ (@legit_api)

업데이트된 Gemini 3 Flash 모델이 LM Arena에 공개되었습니다. LM Arena에서 해당 모델을 이용해 벤치마크·비교 테스트가 가능해졌음을 알리는 짧은 발표입니다.

https://x.com/legit_api/status/2013755037294477439

#gemini #gemini3 #lmarena #llm

ʟᴇɢɪᴛ (@legit_api) on X

An updated Gemini 3 Flash model now available on LM Arena

X (formerly Twitter)

Epsilon (@ElfntOfEpsilon)

LMarena에서 테스트한 모든 변종 모델들이 형편없다고 평가하면서도 웹 개발 영역에서는 사후 학습(post training)이 이루어져 4.1보다 개선된 부분이 있다고 언급. 또한 출시 일정이 조정되어 2월로 연기되었다는 업데이트를 전함.

https://x.com/ElfntOfEpsilon/status/2013259737882583160

#lmarena #modelevaluation #ai #llm

Epsilon (@ElfntOfEpsilon) on X

@daniel_mac8 All variants which have been tested on LMarena are hot garbage. They clearly did a lot of post training on web dev though as it’s a step up from 4.1. Launch has also been pushed to February as they make adjustments, I imagine they are still dissatisfied…

X (formerly Twitter)

Google官方揭AI工具「Nano Banana」命名內幕,靈感來自產品經理的暱稱

julia 2026-01-19 11:16:00 CSTGoogle AI圖像工具「Nano Banana」命名,源於產品經理深夜臨時的戲稱,結合其暱稱「Nano」與「Banana」。此代號因模型廣受歡迎而沿用至今。
https://www.thenewslens.com/article/263741
#LMArena #Nano Banana名稱由來 #Google #Gemini 2.5 Flash Image #科技 #Nano Banana #奈米香蕉 #AI圖像生成與編輯工具 #Google官方部落格 #AI模型 #圖像生成 #命名由來 #生成式AI

Google官方揭AI工具「Nano Banana」命名內幕,靈感來自產品經理的暱稱 - TNL The News Lens 關鍵評論網

Google AI圖像工具「Nano Banana」命名,源於產品經理深夜臨時的戲稱,結合其暱稱「Nano」與「Banana」。此代號因模型廣受歡迎而沿用至今。

TNL The News Lens 關鍵評論網

Markandey Sharma (@TechByMarkandey)

Baidu의 ERNIE-5.0-Preview-1220이 LMArena의 Vision Arena 리더보드에서 1226점을 기록하며 공개되었습니다. 해당 모델은 Vision Arena에서 멀티모달 능력을 평가한 결과 중국 모델 중 1위에 올랐고, 글로벌 톱10에 포함된 유일한 중국 AI 시스템으로 보고되었습니다.

https://x.com/TechByMarkandey/status/2008952454831067594

#baidu #ernie #lmarena #multimodal

Markandey Sharma (@TechByMarkandey) on X

ERNIE-5.0-Preview-1220, from @Baidu_Inc, is now live on LMArena’s Vision Arena leaderboard with a score of 1226. It currently ranks as the top Chinese model on Vision Arena and is the only Chinese AI system in the global Top 10. Vision Arena evaluates multimodal capability,

X (formerly Twitter)
🎭 Behold, the LMArena: the #AI world's favorite fake tan! 🤦‍♂️ Researchers worship this glorified popularity contest, mistaking it for the Holy Grail of AI benchmarks. Meanwhile, it offers all the scientific rigor of a supermarket tabloid. 🥴
https://surgehq.ai/blog/lmarena-is-a-plague-on-ai #LMArena #FakeTan #PopularityContest #AIbenchmarks #ScientificRigor #HackerNews #ngated
LMArena is a cancer on AI

Would you trust a medical system whose only metric was “which doctor wins the Internet?” No, you'd call that malpractice. Yet that's LMArena.

LMArena is a cancer on AI

Would you trust a medical system whose only metric was “which doctor wins the Internet?” No, you'd call that malpractice. Yet that's LMArena.

Minimax-M2.1 vươn lên vị trí #1 mô hình mã nguồn mở trên bảng xếp hạng WebDev và #6 chung cuộc (1445 điểm), ngang bằng với GLM-4.7 trong bản đánh giá mới nhất từ Code Arena. Các mô hình được thử nghiệm qua khả năng tạo website, ứng dụng, trò chơi từ một prompt duy nhất. #AI #Minimax #GLM #WebDev #CodeArena #TríTuệNhânTạo #MãNguồnMở #LMArena

https://www.reddit.com/r/singularity/comments/1pzq0c3/lmarena_minimaxm21_ranks_1_open_model_on_webdev/

GPT-5.2-high xếp hạng 12 trên bảng LMArena, dưới GPT-5.1-high (hạng 6). Dựng bởi Reddit/X. #AI #GPT #LMArena #CôngNghệ #TríTuệNhânTạo #MáyHọc

https://www.reddit.com/r/singularity/comments/1poob4m/gpt52high_scores_12_on_lmarena_underperforming/