RuntimeWire

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

Latest stories — page 10

• Kimi K2.7 ranks second behind Fable 5 and above GPT 5.5 xhigh in ErdosBench's mathematical research test

  Przemek Chojecki's 14-problem smoke run puts Moonshot's new open-weight model behind Claude Fable-5-max and ahead of GPT-5.5 xhigh.

• Depthfirst turns FFmpeg into a proof point for autonomous security agents

  The AI security startup says its agent found 21 FFmpeg zero-days for about $1,000, including an RCE exploit primitive.

• Dapple Raises $30M Seed to Sell Private AI Cloud as Owned Infrastructure

  Founder Tricia Martinez-Saab says Dapple has signed more than $100 million in contracts, but customer names and revenue timing remain undisclosed.

• Head to head: AnimateDiff Turbo vs Luma Ray 3.2 Image to Video

  One model actually stages the prompt; the other mostly vibes around it. Across both tests, Luma Ray 3.2 Image to Video is the only system that consistently delivers the requested setting, motion, and subject choreography.

• Databricks Open Sources Omnigent to Put a "Meta-Harness" Above AI Agents

  Matei Zaharia and Kasey Uhlenhuth's alpha project sits above Claude Code, Codex, the Pi agent and custom agents to compose multi-agent workflows, share live sessions and enforce contextual policies.

• Rio de Janeiro ships an open AI model built on Qwen

  IplanRIO put Rio 3.5 Open 397B on Hugging Face with MIT licensing, a 1M-token context claim, and self-reported gains over Qwen's base model.

• Zuckerberg's AI reorg is messy, unpopular, and probably the job map Big Tech needs

  Meta's 6,500-person Applied AI unit looks less like automation magic than the unglamorous human operating system required to make models useful.

• OpenRouter: Fusion beats DeepSeek-V4-Pro on substance

  Fusion takes the match 34.6 to 32.3 by winning the harder precision tests, while DeepSeek-V4-Pro looks better on presentation and instruction-following in narrower spots. The split is clear: Fusion is the safer model when correctness matters; DeepSeek-V4-Pro is the cleaner stylist when the task is mostly packaging.

• Malware authors use nuclear and biological weapons language to evade scanners

  A Hades supply-chain wave hid weapons-policy bait in non-executing code comments to jam LLM-first malware triage.

• Anthropic says the jailbreak behind Fable 5 shutdown was code review

  The Amodeis' safety-first AI company is now fighting Washington over whether a narrow coding prompt justifies pulling frontier models.

• Z.ai opens GLM-5.2 to every coding-plan tier

  The new flagship adds High and Max reasoning modes and a 1M-context configuration for coding agents such as Claude Code and OpenClaw.

• Head to Head: Claude Fable 5 vs ChatGPT 5.5

  The open-source coding agent says Claude Fable 5 planned better, while GPT-5.5 matched it on execution at lower cost.

• David Sacks Steps Into Anthropic's Fable 5 Export-Control Fight

  Anthropic pulled Fable 5 and Mythos 5 after a June 12 U.S. directive, turning its safety-first model rollout into a policy test case.

• Anthropic shuts off Fable 5 and Mythos 5 after US export-control order

  The directive reaches foreign nationals inside the US, including Anthropic employees, turning frontier-model access into an export-compliance problem.

• Startup Spotlight: Apollo Atomics Is the Nuclear Startup to Watch in the AI Power Panic

  A public roadmap now points to a 2026 compact steam generator demonstrator, 2027 megawatt testing and a 2028 data-center reactor path for the YC-backed nuclear startup.

• Happy Horse routs AnimateDiff Turbo on prompt fidelity

  AnimateDiff Turbo looks slick, but in this matchup it barely showed up for the assignment. Happy Horse won both tests by actually staging the scenes, hitting the objects, and sustaining motion across frames.

• Zyphra Releases ZONOS2, an Open-Weight Real-Time Voice-Cloning Model

  Zyphra is pairing open weights, Apache 2.0 licensing, hosted inference, and its own TTS eval in a direct challenge to closed voice platforms.

• RuntimeWire — Weekly Report Week 2 · June 5 – June 11, 2026

  The June 5-11 spike validated model comparisons as a distribution wedge

• Vercel AI SDK 7 adds HarnessAgent for coding-agent harnesses

  Vercel's changelog says AI SDK 7 adds `HarnessAgent`, an experimental canary API for running Claude Code, Codex and Pi through sandboxed sessions and SDK-compatible streams.

• General-purpose LLMs beat specialized AI tools in Nature Medicine study

  The paper tested OpenEvidence and UpToDate Expert AI against GPT-5.2, Gemini 3.1 Pro and Claude Opus 4.6 across three medical evaluations.

• BitBoard turns AI data chats into rerunnable dashboards

  The startup stores connections, queries and code so AI-generated analysis can be audited and shared instead of buried in chat.

• Juggernaut Flux Lightning edges AuraFlow on image IQ

  AuraFlow steals the poster brief, but Juggernaut Flux Lightning wins the match by being more convincing on the two harder tests: photoreal product realism and moody illustrative storytelling. The margin is slim, yet the verdict is not.

• Scientists engineer CRISPR system that selectively destroys p53-mutant cancer cells

  Researchers led by Jennifer Doudna engineered a CRISPR system that detects mutant p53 cancer cells and triggers their self-destruction, offering a new strategy against one of oncology's most stubborn targets.

• Moonshot's Kimi K2.7 Code lands on Cloudflare Workers AI

  The post gives Moonshot distribution through Cloudflare, but offers no pricing, benchmarks, context length, or model-size details.