Curious on how you can build your own games on Design Arena?
Learn all about our multi-file, multi-turn gamedev arena, where users can build complex single and multiplayer games, with the help of one of our founding engineers @oliverjohansson.
BREAKING - OFFICIAL RESULTS: GPT-5.6 Sol by @OpenAI is 1st overall on Design Arena with an Elo of 1353.
This puts GPT-5.6 Sol above Claude Fable 5 by @AnthropicAI and in the same performance band as GLM 5.2 by @Zai_org on frontend design.
This is an 18-position and 60-point Elo
BREAKING - OFFICIAL RESULTS: GPT-5.6 Sol by @OpenAI is 1st overall on Design Arena with an Elo of 1353.
This puts GPT-5.6 Sol above Claude Fable 5 by @AnthropicAI and in the same performance band as GLM 5.2 by @Zai_org on frontend design.
This is an 18-position and 60-point Elo
Introducing Voice Styles for Text-to-Speech Arena!
Users can now guide model outputs with detailed voice descriptions across dimensions such as tone, emotion, age, accent, and more. This expands the evaluation space beyond audio quality, adding signal on steerability and
Seedream 5.0 Pro by @BytePlusGlobal is 4th overall on Image Arena with an Elo of 1318.
This a 15 rank and 96 Elo point jump over @BytePlusGlobal's second-highest ranked model Seedream Lite 5.0, making Seedream 5.0 Pro their top model. Seedream 5.0 Pro performs especially well in
GPT-5.6 Sol, Terra, and Luna by @OpenAI are now on Design Arena.
GPT-5.6 releases with 3 tiers: Sol, Terra, and Luna. Sol being OpenAI’s flagship model, Terra being a lower-cost option with high-performance, and Luna being the fastest and most cost-efficient model of the line.
Muse Spark 1.1 by @AIatMeta is now available on Design Arena!
Building upon the first Muse Spark, Muse Spark 1.1 features improvements in tool and computer-use workflows, coding tasks, and multimodal understanding with greater cost efficiency.
Congrats to the @AIatMeta team on
We’re excited to introduce Muse Spark 1.1, a significant upgrade from the first Muse Spark model we released earlier this year.
Along with this release, we are launching a public preview of the new Meta Model API where developers can access Muse Spark 1.1.
The model is also
Reve 2.1 by @reve is 2nd overall on Image Editing Arena on Design Arena with an Elo of 1339.
This is 10 Elo points higher than @reve’s previous model, Reve 2.0, closing the gap on GPT Image 2.
With this newest iteration, Reve 2.1 establishes a new Pareto frontier in Preference
BREAKING: Reve 2.1 by @reve is 2nd overall on Image Arena on Design Arena with an Elo of 1350.
Launched only a month after Reve 2.0, Reve 2.1 improves by 17 Elo points, driven by gains in visual intelligence, closing the gap with GPT Image 2 by @OpenAI.
Reve remains the top
Reve 2.1 by @reve showcases enhanced world knowledge and visual reasoning.
See a comparison below of Reve 2.0 and Reve 2.1 generating an image of a grocery store shelf.
On this particular test, Reve 2.1 has the closest price match of the top image models, outcompeting Reve 2.0
BREAKING: Grok 4.5 by @SpaceXAI is 5th on Website Arena with an Elo of 1328.
This is a 25 rank jump over @SpaceXAI’s previous highest performing model.
Trained with Cursor data, Grok 4.5 is in the same performance band as Claude Opus 4.6 (Thinking) by @AnthropicAI, achieving
Announcing Grok 4.5, our first model trained specifically for coding and agents. It was trained with Cursor and offers frontier intelligence at leading speeds and cost efficiency.
x.ai/news/grok-4-5
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
