Let's say your robotic policy fails a task. Your success metrics look fine. What would you actually trust to tell you what went wrong? Poll 21% — 📊 Scalar metrics 12% — 🎥 Rollout video 12% — 🗂️ Failure taxonomy 54% — 🦾 Real-world test 57 votes • 6 days

GPT-5.6 Sol took the top three spots on WolfBench: Codex: 86.74% Terminus-2: 85.17% Hermes: 84.49% So far, so benchmarky. Then I checked the tokens. With the same Sol model at maximum reasoning, Codex used 83.8 million tokens per run. Hermes used 170.9 million. Twice the…

CoreWeave ARIA is our AI Research and Improvement Agent, built into @wandb. Hand it the research when you step away and it reads your runs, forms a hypothesis, launches the next run, and scores it against your baseline. You return to results, its reasoning and next steps.…

So @wandb has our great Bee mascot, @CoreWeave has Arena... why not a Bee Arena? I really wanted to try out Kimi K3 and put it up against GPT-5.6 Sol. GPT5.6 Sol - @OpenAI ------------------------ - Bees are aggressive and swarm you immediately - Lighting leaves you…

Big news, the number keeps on going up