2+ day, 23+ hour ago   (483+ words) DeepSeek V4 Pro, which launched in April 2026 as an open-weight model with 1.6 trillion total parameters and roughly 49 billion active per inference pass, was itself considered unusually compute-efficient for the quality it delivered. Magic's claimed compute for matching it — approximately 1.58 × 10²³ FLOPs — is…...

3+ day, 22+ hour ago   (165+ words) The trajectory — 66 vs. 61, then 57 vs. 55, then 53 vs. 53 — is not evidence that the models converged. No new models were released. The benchmarks changed. Identical composite scores mask two findings that matter more for deployment than the headline number does. The practical…...

1+ week, 3+ day ago   (382+ words) When Qwen3.8-Max launched on August 3, Arena ranked Qwen3.8-Max fourth on the Code Arena: WebDev leaderboard with 1,668 points — trailing Claude Opus 5 (Max) at 1,705, Kimi K3 (Max) at 1,676, and Claude Opus 5 (High) at 1,669. The strength held across categories. Qwen3.8-Max-0902 ranked first in Data…...

4+ week, 1+ day ago   (344+ words) SAEVerbalizer takes a different approach, working from the inside of the model rather than from observations of its behavior. An SAE produces, for each of its learned features, a decoder direction — a vector in the language model's activation space representing…...

1+ week, 4+ day ago   (457+ words) Runway announced that software interfaces no longer have to be built — they can be generated, frame by frame, in real time, with no HTML, CSS, or JavaScript involved. The system is called Solaris, and it represents the company's first Interface…...

2+ day, 23+ hour ago   (524+ words) The result is a global KV cache of 890 bytes per token — roughly one-quarter of V4-Flash and approximately 1/437th of DeepSeek-V1's per-token global KV cache size from just two years ago. The memory savings come from four distinct architectural changes working…...