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 leap from GPT-5.5. GPT-5.6 Sol also establishes a new Pareto frontier for preference vs. speed, faster than any model at this performance. Congratulations to the @OpenAI team on the launch!

First real look at my sword fighting hobby project - Responsive, accurate, freeform combat using physics that are fun, not frustrating, with intuitive controls that are easy to learn, hard to master. A lot of physics based sword fighting games have come out lately but IMO none strike that difficult balance between control, animation and physics...trying my best to pull it off here. This early test is hacked together but it's already showing promise. Feels good to attack and defend and really gives you the feeling that you are fighting with a sword; distance, timing, weapon placement, etc. all matter. Anyway, very WIP with lots to do...and for once, don't look at the feet! I'll take care of that someday ;) #UE5# #gamedev# #physics# #swordfighting#

GPT-5.6 Sol and Luna are ahead of Terra at every point on the Intelligence vs Cost per Task chart. GPT-5.6 Luna stands out as a particularly cost efficient model Charting the Artificial Analysis Intelligence Index shows the trade-off between intelligence and Cost per Intelligence Index Task. Across reasoning efforts, each GPT-5.6 model pushes past GPT-5.5 on the Pareto frontier (excluding non-reasoning). However, Luna and Sol are always ahead of Terra. This means for any Terra effort level, there is a Luna or Sol effort level that is more intelligent at no extra cost, or as intelligent at lower cost.

Symphony’s entire HP rotor is now assembled! Next week it goes to balancing. Normally this is an out-of-house step, but we got our own balancing machine which lives 15ft from the engine assembly cell. So we can do it in a few hours instead of waiting weeks for a supplier

Sol from @OpenAI one-shotted a multi-tentacled creature morphology and RL training harness, launched and monitored the run on @vast_ai on an H200 and created a @threejs + @webgl_webgpu browser environment for the harvested policy to crawl around in. I woke up and this guy was running on my computer

At humans&, we train models from the long-term impacts of their interactions with people. This requires prioritizing long-horizon multi-agent RL. We've developed and are excited to share an open-source, hardware-native 4-bit RL recipe, significantly accelerating training