@Designarena profile picture
Design Arena @Designarena

World's first benchmark for real-world design with 4M creators and counting. Made by @intelligence_ai

Joined June 2025
  • Posts 539
  • Following 10
  • Followers 16,363
313 Photos and videos
Design Arena reposted
Grace Li (ICML 🇰🇷)
@grx_xce
Jul 4
Will you be at ICML? If you’re current obsessions include: • building evals from production data • verifying “non-verifiable” domains • RLHF/post-training/RL(insert letter)F • image/video generation • real-world benchmarks DM me if you’ll be in Seoul! I know a night market with the best tanghulu, and I have some events to invite you to :)
0:01
16
7
126
10,513
BREAKING: Gemini Omni Flash by @GoogleDeepMind is 1st overall on Video Arena with an Elo of 1404. Gemini Omni Flash establishes a 101 point Elo gap over Seedance 2.0 Mini by @BytePlusGlobal in 2nd place, one of the largest leaps we’ve ever seen on Video Arena. This establishes Google as the world’s leading video generation lab, with a leap of 7 positions from their Veo series. Congratulations to the @GoogleDeepMind team on this accomplishment!
63
106
1,020
149,334
BREAKING: Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) by @GoogleDeepMind is 7th on Image Arena with an Elo of 1271. With an average generation time of around 5 seconds, Nano Banana 2 Lite is 37 seconds faster on average than the higher ranking models above it, which establishes a new Pareto frontier in Image Preference vs Speed. Congrats to the @GoogleDeepMind team for this accomplishment!
4
13
149
16,335
Claude Sonnet 5 by @AnthropicAI is now available on Design Arena! Anthropic’s most agentic Sonnet model yet, Claude Sonnet 5 brings major improvements across agent capabilities, coding, and knowledge & professional work, narrowing the performance gap with Opus 4.8 at a lower price point. Congrats to the @AnthropicAI team on the launch!
Claude
@claudeai
Jun 30
Introducing Claude Sonnet 5, our most agentic Sonnet yet. It makes plans, uses tools like browsers and terminals, and runs autonomously at a level that just a few months ago required larger and more expensive models.
0:11
6
6
114
7,217
Mercury 2 by @_inception_ai is now available on Design Arena! Debuting as the world’s first reasoning dLLM (Diffusion Large Language Model), Mercury 2 generates responses through parallel refinement to give users reasoning-grade quality without compromising on speed. Congrats to the @_inception_ai team on the launch!
Stefano Ermon
@StefanoErmon
Feb 24
Mercury 2 is live 🚀🚀 The world’s first reasoning diffusion LLM, delivering 5x faster performance than leading speed-optimized LLMs. Watching the team turn years of research into a real product never gets old, and I’m incredibly proud of what we’ve built. We’re just getting started on what diffusion can do for language.
2:00
2
3
42
2,852
Introducing Video-to-Website on Design Arena! You can now generate websites from video and text inputs on Design Arena, making it easier to create dynamic, high-fidelity sites. Leaderboard coming soon!
0:04
3
2
37
2,370
Try it now at DesignArena.ai

Design Arena

Design Arena is the largest global crowdsourced benchmark for design. Challenge, Vote, Crown your Winner.

designarena.ai
7
668
BREAKING: LTX-2.3 by @ltx_io is 1st overall for open weights on Video Arena with an Elo of 1138. This makes LTX-2.3 their top video generation model, with a 115 Elo jump over the LTX’s second-highest ranked model, LTX-2 19B, currently in 5th with an Elo of 1023. Congrats to the @ltx_io team for the achievement!
9
24
269
32,646
We’re excited to be hosting our first office-warming tomorrow! If you’re an eval enthusiast in the SF area, and you want to meet more of the community, we’d love to see you there. luma.com/lpob09l0

The Intelligence Company Office Warming · Luma

Sooooo.... we moved out of our apartment :) Less than a year ago, we started with a simple belief: the way we evaluate AI will shape what AI becomes good at.…

luma.com
4
6
32
10,791
Cosmos 3 Super by @NVIDIAAI is now available on Design Arena! Cosmos 3 is the world’s first open weight omnimodel with native vision reasoning and multimodal generation capabilities across text, image, video, and audio. Part of the Cosmos 3 family, Cosmos 3 Super is a 64-billion parameter model capable of creating high-fidelity images from text inputs and dynamic, high-quality videos from image inputs. Congrats to the @NVIDIAAI team on the launch!
1
5
69
4,020
An image generated by Cosmos 3 Super by @NVIDIAAI to the prompt: "peaceful mountain lake at sunrise with lily pads, pink water lilies, reflective water, and misty forested mountains"
8
822
Seedance 2.0 Mini by @BytePlusGlobal is now 1st on Video Arena on Design Arena with an Elo of 1342. With @BytePlusGlobal models holding the top 3 in Video Arena, they further establish their position as the top lab for video generation, followed by @HappyHorseATH and @xAI. Compared to the flagship Seedance 2.0 model, Seedance 2.0 Mini achieves a 13-point Elo improvement. Congratulations to the @BytePlusGlobal team on the achievement!
6
16
198
11,071
Comparison of Seedance 2.0 and Seedance 2.0 Mini generating a video from the prompt: “Cinematic close-up from behind a racecar driver”
0:05
1
13
1,219
GLM-5.2 by @Zai_org is 5th in Mobile App Arena on Design Arena with an Elo of 1248. This is a 2 position jump from GLM-5.1, putting GLM-5.2 in the same performance band as Claude Sonnet 4.6 by @AnthropicAI. @Zai_org is the top open-weight lab in Mobile App Arena and the third lab overall, following @GoogleDeepMind and @AnthropicAI. Congratulations to the @Zai_org team on this accomplishment!
2
9
122
7,207
Nemotron 3 Ultra establishes @NVIDIAAI as the top 9 open-weight lab on 3D Design Arena. With an average input cost of $0.50, the model achieves strong performance while remaining one of the most cost-efficient in its class. This puts NVIDIA among the leading open-weight labs in 3D Design Arena, following closely behind @Alibaba_Qwen. Congratulations to @NVIDIAAI on the achievement!
1
2
45
3,480
Design Arena’s benchmarks now help power @OpenRouter’s MCP Get live model intelligence directly in your agent!
OpenRouter
@OpenRouter
Jun 25
Replying to @OpenRouter
The model performance rankings come from our new Benchmarks API, allowing your agent to query live benchmark scores (incl Artificial Analysis and Design Arena) Fun result: @Zai_org’s GLM-5.2 is the best available model for both coding & design Docs: openrouter.ai/docs/api/api-r…
1
6
31
6,397
HappyHorse 1.1 by @AlibabaGroup is 4th overall on Video Arena with an Elo of 1320. This puts it in the same performance band as Seedance 2.0 Fast by @BytePlusGlobal. HappyHorse 1.1 is now the top model by @AlibabaGroup, leading with a 34 Elo jump over their previous model HappyHorse 1.0, currently in 6th. Congrats to the @HappyHorseATH team on the launch!
12
5
90
5,322
Comparison of HappyHorse 1.0 and HappyHorse 1.1 by @AlibabaGroup generating a video from the prompt: "A spaceship flying towards Earth.”
0:05
1
8
841
Seedance 2.0 Mini by @BytePlusGlobal is now available on Design Arena! A lower-cost version of Seedance 2.0, Seedance 2.0 Mini generates videos at around twice the speed with visual consistency across multi-shot scenes. Congrats to the @BytePlusGlobal team on the launch!
1
1
17
1,472
Nex-N2-Pro by @NexEcosystem is 12th among open weights on Design Arena with an Elo of 1275. This is in the same performance band as MiniMax M2.7 by @MiniMax_AI and GLM 4.7 by @Zai_org. Nex-N2-Pro performs strongest in 3D Design Arena, where it ranks 16th with an Elo of 1314. Congratulations to the @NexEcosystem team on the achievement!
1
2
41
4,757
Load more