Summary
A thin evening slot with no curated @bayesiansapien retweets and most of the feed dominated by French politics from @brivael and LA mayoral content from @spencerpratt. The one piece of hard AI news is @ns123abc's claim that xAI's Grok Foundation Model V9-Medium (1.5T parameters) has finished training and drops in two weeks, a 3x jump over the 0.5T V8-small that currently serves Grok traffic. Adjacent to that, two cross-source Grok-Build signals show up: @kilocode's coding-gateway usage numbers (Stepfun still leading at 30%+ but Grok Build 0.1, Gemini 3.5 Flash, and Qwen 3.7 Max are reshuffling the rankings), and a @brivael repost of a coder claiming Grok Build is "right there with Claude" on agentic coding. Off to the side, a French repost surfaces Pope Leo XIV saying the Vatican and Anthropic will collaborate on "a path for humanity in the AI era" and that AI must be "disarmed" like nuclear energy. Everything else is noise.
Posts
- Grok V9-Medium (1.5T) finished training, drops in 2 weeks (@ns123abc). Quoted internal text: "This will be a major improvement over the 0.5T v8-small that currently serves all Grok production traffic, especially for difficult coding tasks." 3x scale jump on the production model. If real, this is the most consequential xAI signal of the week. Follow-up tweet from the same account ("ITS HAPPENING") appears to be the same cluster.
- Kilo Gateway usage: three challengers to Stepfun (@kilocode). Stepfun still 30%+ of usage on the Kilo coding gateway, but Grok Build 0.1, Gemini 3.5 Flash, and Qwen 3.7 Max are now in the mix on the same workflows and prompts. Useful market-share data point on what coders are actually picking when given a router across frontier models. Qwen 3.7 Max ties to the 05-23 autonomous-coding launch.
- Grok Build "right there with Claude" on real coding (@brivael, repost of @scion_x_). One developer's hands-on report after a full agentic coding session: model is sharp, agentic flow holds on complex tasks, has personality, some UX rough edges. Single-anecdote signal but it pairs with the Kilo numbers above. Two independent data points in the same slot is the first time Grok Build has registered as a serious Claude competitor in the wiki.
- Pope Leo XIV: Vatican and Anthropic to work together on "a path for humanity in the AI era" (@brivael, repost of French source). Quoted: AI must be "disarmed", compared to nuclear energy. No primary English source attached, video only. Adds to the pattern of Anthropic doing institutional alignment work outside the standard tech-policy channels.
- @brivael French politics and lifestyle cluster (cluster of 14) (@brivael). XChat homescreen post, French presidential satire, Attal video commentary, Elon-as-French-president fiction, business school as ponzi, Yalc.ai promo, coffee machine post, Musk wealth math, UAE growth, and similar. None research-relevant. Skip.
- @spencerpratt LA mayoral campaign (cluster of 4) (@spencerpratt). Karen Bass homelessness clip, NBC News attention-vs-votes piece, LA Times crime-rate framing, Drew Carey response. Caught by the keyword filter, not AI-relevant. Skip.
- @AustinJustice Travis County storage burglar (@AustinJustice). Local Austin crime story. Skip.