Moonshot ships Kimi K3, a 2.8T-parameter model with 1M-token context [Xtra!]A massive open contender pressures Western labs on scale and context length.
Fireworks hits $17.5B valuation on cheap-inference demand [Xtra!]Signals a market shift toward cost-optimized inference infra rather than chasing the newest frontier model.
NVIDIA's Nemotron 3 Embed tops retrieval benchmark [Xtra!]Embedding models quietly underpin every RAG and search stack.
Grok's CLI caught uploading local files to the cloud [Xtra!]A stark reminder that agentic dev tools can leak data without users realizing it.
Why agents that pass tests fail in production [Xtra!]Passing evals isn't the same as being production-ready.
"Harness Handbook" maps agent behavior to code [Xtra!]Harness engineering - the scaffolding around a model - is becoming its own discipline.