Alibaba's Qwen team has a running habit of shipping models that make you question your assumptions about parameter counts.
For two months, a model called "Owl Alpha" quietly climbed the OpenRouter charts. No launch event, no press tour, no Twitter hype cycle.
Two days ago at Build, Microsoft did something it's never done before: announced a full family of homegrown AI models that compete directly with the...
Two days ago MiniMax dropped a launch announcement for M3 that reads like a greatest-hits compilation of everything the open-weight community has been asking...
The biggest announcement from Microsoft Build 2026 wasn't a new Azure service or another Windows agent feature. It was a quiet decoupling.
Poolside raised $626 million, hired a team of ex-DeepMind and ex-Meta researchers, and then went quiet for nearly three years.
Zyphra just dropped a model that makes me question everything I assumed about parameter efficiency. ZAYA1-8B is an 8.
Every large language model you've touched — GPT, Claude, Llama, Qwen, Gemma — generates text the same fundamental way. Predict the next token.
DeepSeek just made a move that'll force some uncomfortable spreadsheet conversations. The 75% promotional discount on V4 Pro?
A model with 760 million active parameters just scored 91.9% on AIME 2025 with test-time compute.
Moonshot AI's Kimi K2.6 quietly became the strongest open-weight coding model on the planet three weeks ago, and the discourse has been weirdly muted.
NVIDIA quietly shipped one of the most interesting open model architectures I've seen this year, and most of the coverage buried the lede.
DeepSeek V4-Pro grabbed headlines three weeks ago with an 80.
Most open models fight for the same throne: highest MMLU score, best SWE-Bench pass rate, flashiest reasoning demo.
On May 3rd, Moonshot AI's Kimi K2.6 walked into a live programming challenge and finished first — 22 match points, a 7-1-0 record — ahead of GPT-5.
Most open model launches follow a predictable script: bigger parameters, higher benchmark scores, a claim about beating GPT-something on MMLU.
Every multimodal agent demo I've seen follows the same script: chain a vision model to a language model, bolt on Whisper for audio, pray the latency...
Nine days ago Moonshot AI dropped Kimi K2.6 — a trillion-parameter MoE model that beats Claude Opus 4.