RuntimeWire

Real-time startup intelligence for the AI economy. Funding, model launches, infra shifts, and founder moves — the signals that matter.

Latest stories

• SoundCloud traded frequencies above 17 kHz for cleaner AAC streams

  The 2025 encoder swap made spectrograms look worse at 256 kbps, while SoundCloud left its testing methodology undisclosed.

• Your Fair Share turns SEC filings into a profit calculator for employees

  The anonymous builder behind WorkerOwned.info says the free tool covers about 940 public companies, with roughly 200 manually checked.

• Alibaba puts Qwen3.8 preview in paid products before open-weight release

  The model is available inside Alibaba products, while architecture details, benchmarks, licensing terms and a weight-release date remain undisclosed.

• Stripe and Advent bid $53 billion for PayPal during Lores turnaround

  PayPal's new CEO must weigh a months-old operating reset against a $60.50-a-share bid that directors already view as inadequate.

• Baidu open-sources 3B OCR model for one-pass, multi-page PDF parsing

  Unlimited-OCR keeps its decoder cache fixed, though the viral 100-page claim runs ahead of Baidu's published tests.

• Moonshot pauses Kimi subscriptions as K3 demand strains GPU capacity

  The AI developer will protect existing users, add compute and separate its coding plan from its general-purpose membership.

• Current AI launches Alpha Chat to assemble a public open-source AI stack

  CEO Ayah Bdeir is applying her littleBits playbook to a nonprofit coalition backed by more than EUR400 million in commitments.

• Harvey open-sources 10 M&A diligence environments to train legal agents

  The synthetic data rooms turn deal review into reinforcement-learning infrastructure, though Harvey has yet to publish agent results.

• Fireworks raises $1.505B as Lin Qiao scales open-model inference

  The former PyTorch leader is valued at $17.5 billion on a bet that companies will customize their models while outsourcing the machinery underneath.

• Northwestern's Phantom Twist spins at 1,500 RPM to evade human vision

  Michael Rubenstein's lab used computational design to make a single-propeller aircraft fade into its background, though it remains confined to controlled flights.

• Moonshot's Kimi K3 jolts AI stocks before its weights are public

  Yang Zhilin's 2.8 trillion-parameter model topped a coding leaderboard, while Moonshot concedes it still trails the leading US systems overall.

• Zoox recalls 105 robotaxis after one entered a smoke-covered fire scene

  The software fix tests Jesse Levinson's ground-up autonomy bet as regulators press robotaxi operators to handle emergencies like human drivers.

• REO Trucks draws 12,173 reservations for a $21,500 pickup without a prototype

  Founder Zach De Bernardi is using demand for a simple gas truck to recruit the suppliers, engineers and capital needed to build it.

• Startup Spotlight: Novo Space builds modular computers for satellite constellations

  Co-founders Rodrigo Diez and Facundo Jorge are building modular flight computers designed to reduce repeated engineering work across satellite missions.

• Handy creator released transcribe.cpp to unify local speech-to-text models

  CJ Pais turned Handy's cross-platform deployment problems into an MIT-licensed engine backed by Mozilla.ai's residency program.

• Head to head: Google: Gemini 3.1 Flash TTS Preview vs Eleven v3

  This matchup turned on delivery, not diction. The two models were even on tricky pronunciation, but Gemini 3.1 Flash TTS Preview pulled ahead where performance mattered most: believable news reading and more convincing whisper dynamics.

• Head to head: AuraFlow vs Ideogram V4.0 Text to Image

  One model consistently followed the brief; the other too often wandered off it. In a matchup decided by prompt discipline rather than vibes, the result wasn’t remotely close.

• Cribl adds CardinalOps' Israeli security team in estimated $100 million deal

  The deal adds about 20 security engineers and detection technology as Cribl builds toward an open alternative to legacy SIEM platforms.

• Head to head: Kimi K3 vs xAI: Grok 4.5

  This was a thin matchup on paper, but Kimi K3 still comes out ahead on the numbers. The catch: the only judged task showed enough order sensitivity that this result reads as a lean, not a rout.

• Berkeley researcher used GPT-5.6 to derive a Lean-verified optimization bound

  Phillip Kerger's preprint narrows a gap dating to 1996, though the paper remains unreviewed and Lean covers only its lower-bound result.

• Z.ai hits $1 billion annualized sales pace as its coding bet accelerates

  The partial-month figure is far ahead of Z.ai's 2025 revenue, while contract mix, margins and durability remain unresolved.

• LinkedIn scripts 80% of its agent workflow to limit hallucinations

  Walmart is governing employee-built agents, while Forethought co-founder Sami Ghoche says Zendesk's 20 billion conversations still need data pipelines.

• DeepSeek API tests show Claude-like behavior under selected prompts

  The results revive Anthropic's distillation claims, though the test cannot establish who generated the outputs or whether they were used for training.

• DeweyLearn raises $5 million to grade real-world skills with multimodal AI

  SJF Ventures led the Series A for Luyen Chou and Dirk Liebich's assessment startup, which is already grading culinary and clinical training work.