Articles
A curated summary of the most important AI developments each week.
Grok 4.5 Is Not "Opus-Class." It Is Something More Interesting: Opus-Adjacent and Much Cheaper.
July 10, 2026
Grok 4.5 trades outright Opus dominance for strong coding performance, lower prices, and a token-efficiency advantage shaped by Cursor.
Claude Sonnet 5 Finally Launched. The Rumors Were Wrong About Almost Everything Except That It Was Coming.
June 30, 2026
Claude Sonnet 5 brings near-Opus agent performance at lower cost. Its official benchmarks show how far months of launch rumors missed.
The US Government Is Now Approving AI Models Before They Ship. The Question Nobody Is Asking Is Why Not Fix the Software Instead.
June 27, 2026
The US is vetting frontier AI releases for cyber risk. The harder question is whether policy should gate models or fix vulnerable software.
Sakana Fugu Beats GPT-5.5 With Model Orchestration
June 25, 2026
Sakana Fugu shows a new AI architecture: learned orchestration that routes tasks across frontier models instead of training one bigger model.
The Real AI Race Nobody Is Covering: MiniMax M3, GLM 5.2, and DeepSeek V4 Are Fighting Each Other
June 18, 2026
MiniMax M3, GLM 5.2, and DeepSeek V4 show the real open-weight AI race: coding, scale, price, and developer ecosystems.
Claude Fable 5 Is the Mythos Model You Can Actually Use. But there's a Catch.
June 9, 2026
Claude Fable 5 brings Mythos-class capability to public Claude users, but safeguards, pricing, and missing DeepSWE scores complicate the launch.
Ideogram 4 Open Weight Release: The Image Generation Gap Just Closed on Hugging Face
June 3, 2026
Ideogram 4 brings open-weight image generation close to the closed frontier with 9.3B params, strong text rendering, and Hugging Face weights.
MiniMax M3 Is the First Open-Weight Model With Frontier Coding, 1M Context, and Multimodality. The Weights Aren't Out Yet.
June 2, 2026
MiniMax M3 claims frontier coding, 1M context, and native multimodality as open weights, but the weights are not released yet.
NVIDIA RTX Spark Is Not a Chip. It's the End of the Cloud-First AI Era.
June 2, 2026
NVIDIA RTX Spark brings petaflop AI, 128GB unified memory, and Windows-native agents to local PCs, challenging cloud-first AI.
Claude Opus 4.8 Released: Dynamic Workflows, Cheaper Fast Mode, and a Missing Benchmark
May 28, 2026
Claude Opus 4.8 improves coding, honesty, fast mode, and dynamic workflows, but launches without a DeepSWE score.