GPT-5.6 Sol Builds Cannon in Blender at 750 Tokens per Second
The system used GUI controls instead of Blender's Python API.
A short X post from @evayzh showing GPT-Sol building a cannon model inside Blender is getting attention because the clip appears to run at normal speed, not sped up. In a post on X, @kimmonismus called the demo "so much more impressive than looking at benchmarks," tying that reaction to a claimed 750 tokens per second. A later follow-up post on X said a correction made the demo look "even more impressive."
Combined views
64.8K
GPT-5.6 Sol Builds Cannon in Blender at 750 Tokens per Second
The system used GUI controls instead of Blender's Python API.
A short X post from @evayzh showing GPT-Sol building a cannon model inside Blender is getting attention because the clip appears to run at normal speed, not sped up. In a post on X, @kimmonismus called the demo "so much more impressive than looking at benchmarks," tying that reaction to a claimed 750 tokens per second. A later said a correction made the demo look "even more impressive."