World Wire

A stock voice only needs to sound good. A cloned voice must sound like one specific person. That is a harder test, and the 7 Voice cloning providers here handle it very differently. We recorded one 10-second reference clip and cloned it on every platform. The results sit side by side below, unlabeled, so you can judge before seeing names. We then compare 4 axes that decide deployments: reference audio needed, consent verification, commercial license, and languages. How the 10-second test works The reference is 10 seconds of one speaker in a quiet room, saved as WAV. Every platform reads the same 2 sentences. One is conversational; the other packs numbers and a proper name. We use each vendor’s instant cloning path with default settings. Two vendors bend the rule. Hume documents 15 seconds as its floor, so its clone uses the same speaker extended to 15 seconds. ElevenLabs recommends 1 to 2 minutes for Instant Voice Cloning, so its 10-second clone runs below vendo...

StepFun has released Step 5 Preview , its new flagship model for agentic work. The target workloads are software engineering, professional knowledge work, and finance. The main pitch is cost. StepFun team states the model delivers comparable intelligence at a substantially lower task cost. That is the ‘Pareto frontier’ framing in the launch title. Is it deployable? Yes, as a hosted API and on the StepFun platform . Self-hosting waits for open weights. StepFun says open weights land on October 15, 2026. By simple arithmetic, 600B parameters need about 1.2 TB in BF16, before KV cache. Plan for multi-GPU server hardware once weights ship. What StepFun Shipped Step 5 Preview is a sparse Mixture-of-Experts (MoE) model. It holds about 600B total parameters and activates about 27B per token. That is roughly 4.5% of the weights per token. The official model documentation lists these specs: Model ID: step-5-preview Context window: 1M tokens Input: text...

Google confirmed on Friday, September 18, 2026 that a Gemini model accessed 3 outside companies’ systems. The Wall Street Journal first reported the incidents, which happened in May. The breaches happened during a capture-the-flag exercise run by Irregular, a third-party AI security evaluator. Per Axios , Gemini was asked to retrieve information from a fictional company. That fictional company shared its name with a real one. The test was never supposed to touch the internet. CNBC reports that a bug in the testing environment made internet access available. The techniques were basic. In 1 case, Gemini guessed passwords until it got in. In the other 2, it used credentials found in a public repository. Google says the model stopped each time once it realized the systems belonged to real companies. Heather Adkins, Google’s VP of security engineering, said in a statement reported by CNN that the 3 entities were made aware, and that Google worked with its trai...