Google's Gemini 3.8 Flash is live and callable through the Gemini API, with RuntimeWire testing it head-to-head against the Fable 5.1 model, though Google's public documentation still lists Gemini 3.7…
RuntimeWire opened a private beta for Gateway on September 1st, a product that lets companies govern AI models used through employees' existing AI clients and coding tools without forcing a new interf…
RuntimeWire, a tech news outlet, reports that its AI sidekick model, which automates routine newsroom tasks like embargo calendars and transcripts, enabled the publication to publish more than 80 stor…
RuntimeWire launched a shared Grok Bot on August 29th that briefs users on AI startups, funding rounds, and founder moves, pulling stories from its news feed and checking primary sources before answer…
OpenAI has built an unannounced, gated review system in its ChatGPT/Codex desktop client that lets users attach native Office comments or private ChatGPT assignments to specific Excel ranges and Power…
Kimi Desktop 3.2.3 contains an unannounced 'Selection toolbar' that detects highlighted text systemwide and displays AI Search, Translate, Summary, Copy, and Read Aloud buttons, but the actions are no…
OpenAI is building a dormant "Subscription sharing" feature, internally called ChatPass, discovered in the Codex desktop client by RuntimeWire, which could let third-party AI applications draw from a …
RuntimeWire reverse-engineered OpenAI's Codex desktop client and found hidden mini-games—Asteroids, Snake, and Brick Breaker—embedded in production modules named codex-micro-mini-games and codex-micro…
OpenAI's Codex desktop client contains an undocumented GenUI architecture that can refresh interfaces inside ChatGPT conversations, with a bundled catalog of 467 versioned 'Learning Block' types spann…
Claude Opus 5 (Fast) topped the Newsroom Reliability v0.2 benchmark with a mean score of 0.73 at an estimated $0.0460 per task, according to a 50-task run by RuntimeWire. Z.ai's GLM 5.3 Flash, Google'…
Z.ai's GLM 5.3 Flash ranked first in a 12-task Editorial Craft benchmark with a mean score of 0.94 and an estimated cost of $0.0002 per task, according to RuntimeWire. Step-3.7-flash followed at 0.92,…
Imageat, a third-party API provider, is serving an unidentified uncensored derivative of Alibaba's Qwen3.8-27B model without disclosing the checkpoint, revision, modification method, or endpoint evalu…
Google Search impressions for AI-native newsroom RuntimeWire fell roughly 97% in five days, from 42,237 on August 12th to 1,327 on August 17th, following a rollout of translated pages, and the English…
Metricform, founded by Lindsey Aliksanyan, launched an AI-powered narrative intelligence platform on August 25 that monitors news, social media, forums, reviews, podcasts, the open web, and LLM-genera…
OpenAI is testing an undocumented fallback tier called Luna Reserve for its Codex coding agent, which switches eligible Plus and Pro users to a separate gpt-reserve model after their advanced-model al…
Replit launched Free Mode, a lower-cost setting for its coding agent powered by OpenAI's GPT-5.6 Luna, giving paid Core and Pro subscribers credit-free everyday tasks with five-hour limits and up to 3…
Anthropic has built an unreleased, Mac-first meeting recorder inside Claude Desktop, codenamed Parka, that can turn meeting transcripts into Cowork, Claude Code, or manual follow-up actions, according…
Cursor 3.16.17 contains a hidden, gated Desktop Bridge that lets the bundled cursor CLI enumerate agent threads in Cursor Desktop and submit follow-up instructions, as verified by RuntimeWire through …
An AI-run news outlet, RuntimeWire, beat WIRED by more than three hours in reporting on OpenAI's rogue AI agents talk at the Black Hat security conference, publishing about six minutes after the sessi…
SpaceXAI's updated acceptable use policy, effective August 14, 2026, bars customers from using its services or outputs to help build products that compete with SpaceXAI 'directly or indirectly,' a bro…