Perplexity cuts Decision API pricing to 2 cents per million input tokens

Musubi's open-weights decision model applies plain-English content policies to messages in under 50 milliseconds, without retraining when policies change. Co-founder Filip Jankovic says it lets platforms label content proactively as volume grows.

A user with three months of AI image generation experience posted a Krea 2 workflow built on a character LoRA, tweaked from Koo's Discord workflow. It runs on an RTX 5060 8GB with 16GB DDR3 RAM.

Together AI is the first customer on a dedicated NVIDIA B300 GPU inference cluster on IBM Cloud, using Spectrum-X Ethernet networking. Together says it serves hundreds of trillions of tokens per month to over a million developers.

A developer ran Qwen3.8-Flash-Next locally on an ASUS ROG Strix Halo laptop against Claude Opus 5.5 on the same high-complexity feature in LlamaStash, a large Rust project. The comparison covers weeks of local coding work.

Verified security professionals get reduced blocking classifiers and access to Claude Opus 5.5, Sonnet 5.5, and Mythos 5.1. The program merges Project Glasswing and the original CVP, which ran for six months.