Skip to content

xAI / Grok

Everything xAI and Grok: model iterations, compute build-out and integration with X.

Latest picks

41–60 of 108

Jul 9Thursday

AI HOT (Curated Pool)

Lawsuit: Man used Grok to make 7K sex images of stepdaughter, then shot himself

A new lawsuit alleges xAI's Grok was used to create over 7,000 child sexual abuse images of the user's stepdaughter. The man later shot himself. xAI reported only one gang-rape prompt to NCMEC and did not report the thousands of other CSAM generations. The suit accuses X and xAI of shielding child predators. The post does not spell out why xAI's safety filters missed the bulk of the images, nor whether Grok's image generation disables real-face simulation by default.

Why it matters: A lawsuit says xAI reported only one of more than 7,000 prompts used to generate sexual images of a man's stepdaughter, putting its safety reporting under court scrutiny.

Jul 1Wednesday

AI HOT (Curated Pool)

xAI launches Voice Agent Builder beta: build a voice agent in under 2 minutes

xAI packaged Grok Voice into a no-code platform, now in beta as of July 1. You describe the call flow in plain language, upload docs as a knowledge base, and connect tools like calendars or ticketing systems—then you get a working voice agent. It uses a speech-to-speech path instead of chaining ASR→LLM→TTS, which xAI claims cuts latency and failure points. Pricing is $0.05/min of audio plus $0.01/min for a platform-provided number. xAI also published τ-voice Bench scores: Grok Voice Think Fast 1.0 hit 67.3% overall, versus 43.8% for Gemini 3.1 Flash Live and 35.3% for GPT Realtime 1.5. Take the benchmark with a grain of salt—it's xAI's own test, and third-party results aren't out yet.

Why it matters: xAI turns Grok Voice into a no-code platform where users describe call flows in natural language, a speech-to-speech route that skips the usual three-stage pipeline.

Jun 28Sunday

AI HOT (Curated Pool)

Grok 4.5 enters private testing at SpaceX and Tesla, performance near Opus

Elon Musk says Grok 4.5 is built on a 1.5T-parameter V9 base model with Cursor data added during supplementary training, now in private testing at SpaceX and Tesla. Early evals show performance close to or possibly exceeding Opus. RL is still improving the model, and the Grok Build toolchain is maturing. SpaceX will also release a fully from-scratch trained model every month this year. The post doesn't specify which Opus model, benchmarks, or testing scale.

Why it matters: Musk says Grok 4.5 is already being tested inside SpaceX and Tesla, built on a 1.5-trillion-parameter V9 base model with Cursor data in training, and its performance is initially close to Opus.

AI HOT (Curated Pool)

SpaceX files SpaceXAI trademark, Musk says xAI will merge into SpaceX

SpaceX has filed a trademark for SpaceXAI. Musk says xAI will dissolve as a standalone company and become SpaceX's AI product line. The post doesn't disclose a merger timeline, team structure, or what happens to existing xAI products like Grok.

Why it matters: Musk says xAI will fold into SpaceX as an AI product line, and the SpaceXAI trademark is already filed, showing how an independent lab becomes an internal division.

Jun 24Wednesday

Hacker News front page

Reid Hoffman: SpaceX is 'not an AI company,' xAI is a 'complete train wreck'

LinkedIn co-founder Reid Hoffman called SpaceX 'not an AI company' and xAI 'a complete train wreck' on a podcast. He said SpaceX's post-IPO Cursor acquisition is buying relevance, and its compute leasing is just 'a premium-priced CoreWeave.' On xAI, all 11 co-founders have left and the company is on its third restart. Hoffman also criticized the U.S. government's forced takedown of Anthropic's Fable and Mythos models as 'autocratic willy-nilly,' troubled by the asymmetry with OpenAI. He invests in both Anthropic and OpenAI and sees room for both to win.

Why it matters: Reid Hoffman gives a first-hand read on AI camp lines, calling SpaceX's Cursor deal market-cap posturing and xAI a third rebuild after all eleven co-founders left.

Jun 22Monday

AI HOT (Curated Pool)

Grok Build adds /goal mode for long-running autonomous task execution

xAI added /goal to Grok Build: give the agent an objective and it plans, breaks work into a checklist, and executes until done. You can check status, pause, resume, or clear the goal mid-run. The post doesn't disclose max run time, resource costs, or specific pricing.

Why it matters: Grok Build's /goal command pushes xAI toward the long-running autonomous agent pattern Cursor and Claude Code already use.

Jun 19Friday

Latent Space

Anjney Midha on AI compute waste: frontier labs run sub-10% MFU, AMP plans an independent compute grid

Anjney Midha discusses hidden AI infrastructure waste on Latent Space. xAI's training MFU is under 10%, while Google treated 95% utilization as an outage; best-in-class today is 60–70%. He invested in Anthropic, Mistral, and Black Forest Labs, and now runs AMP, aiming for a 1.2 GW base-load compute grid with 6 GW spike capacity. He also flags DeepMind's unpublished research as a market failure and notes Anthropic prioritized coding as P0 from day one. The post does not disclose a timeline for AMP's grid.

Why it matters: Anjney Midha puts xAI's training compute utilization below 10%, against Google's 95% failure threshold and a top-tier 60-70%, making waste measurable.

Jun 18Thursday

Hacker News front page

OpenRouter ran 11 LLMs in a 30-game battle royale — Grok 4.1 Fast won 43%

OpenRouter's Jacky Liang dropped 11 LLMs into a 2D battle royale for 30 matches. Grok 4.1 Fast won 13 games at $0.97 per win; Claude Sonnet 4.6 won 5 at $26.78 per win — a 27x gap. GPT 5.4 had the most kills (38) but only 2 wins, so killing more didn't mean winning more. GPT 5.4-mini, DeepSeek 4 Flash, and Kimi K2.6 spent $57 combined and won zero games. The models reasoned, called tools, and updated memory each turn — they weren't just generating control code. The post doesn't provide the full leaderboard or detailed behavioral differences across all models.

Why it matters: OpenRouter's Jacky Liang ran 11 models through 30 rounds of a 2D battle royale: Grok 4.1 Fast won 13 games at $0.97 per win, Claude Sonnet 4.6 won 5 at $26.78.

Jun 16Tuesday

AI HOT (Curated Pool)

DOJ invokes national security to defend xAI's unpermitted gas turbines in NAACP lawsuit

The DOJ moved to dismiss an NAACP lawsuit against xAI, arguing that shutting off its gas turbines would threaten military operations. A DOD official stated Grok is one of four models supporting mission-critical work on classified networks, including recent strikes on Iran. The NAACP sued because xAI runs unpermitted turbines at its Colossus 2 site in Mississippi—turbine count grew from 27 to 57 since April, with a 111% spike in nitrogen oxide emissions. The post doesn't specify which national security statute the DOJ is citing.

Why it matters: The DOJ argues shutting down xAI's gas turbines would threaten military operations, weighing Colossus 2's unpermitted emissions against defense needs.

AI HOT (Curated Pool)

SpaceX to acquire AI coding startup Cursor for $60B in stock, days after its IPO

Days after its historic IPO, SpaceX agreed to buy AI coding startup Cursor for $60 billion in stock. Cursor was about to close a $2B round at a $50B valuation from a16z, Thrive, and Nvidia. SpaceX told IPO investors its AI addressable market is $26 trillion and wants the deal to help its xAI-built AI unit catch up with major labs. The transaction is expected to close in Q3. The post doesn't spell out product integration plans, team retention, or regulatory approvals.

Why it matters: A $60 billion all-stock deal for Cursor shows SpaceX using equity to buy AI assets right after its IPO.

AI HOT (Curated Pool)

xAI launches Grok Imagine Video 1.5: faster image-to-video with synced audio

xAI upgraded its image-to-video model to 1.5, now generally available via API with a Fast variant on Grok and mobile apps. A 6-second 720p clip takes about 25 seconds, nearly twice as fast as the previous 40+ seconds. Audio is generated in the same pass—ambience, effects, and dialogue land on the action with better lip sync. Motion holds up over longer clips with fewer warps and more believable weight. Three workflow features are rolling out: Projects for organization, parallel multi-agent prompting, and library search. The post doesn't disclose training data scale or pricing.

Why it matters: xAI upgraded its image-to-video model to 1.5 and opened it through an API, generating a 6-second 720p clip in about 25 seconds with synced audio and lip movement.

Jun 15Monday

AI HOT (Curated Pool)

Grok Build adds Agent Dashboard to manage multiple coding sessions at once

xAI shipped a terminal dashboard for Grok Build that lets you monitor and interact with multiple coding sessions on one screen. Sessions are grouped by state—blocked ones rise to the top—so you handle approvals and questions inline without switching contexts. You can peek at output, reply, dispatch new work, and jump into any session. Closing the dashboard leaves everything running; reopening it restores all sessions. Install via curl -fsSL https://x.ai/cli/install.sh | bash, then run grok dashboard or hit Ctrl+\.

Why it matters: xAI added a terminal control panel to Grok Build that monitors and drives multiple coding sessions on one screen, pinning sessions stuck waiting for a reply, useful for anyone running coding tasks in parallel.

Jun 11Thursday

AI HOT (Curated Pool)

xAI launches a built-in plugin marketplace for Grok Build, starting with 6 partners including MongoDB and Vercel

xAI added a plugin marketplace directly inside Grok Build, so you can browse, install, and update plugins without leaving the terminal. Each plugin bundles skills, slash commands, agents, hooks, MCP servers, and LSPs. The launch lineup includes MongoDB, Vercel, Sentry, Chrome DevTools, Cloudflare, and Superpowers. Every remote plugin is pinned to a specific commit SHA and verified at install time. Developers can publish their own by submitting a PR to xai-org/plugin-marketplace. The post doesn't mention any review process or revenue split.

Why it matters: xAI bundles skills, slash commands, agents, hooks, MCP servers and language servers into one plugin format for its CLI, with six launch partners.

Jun 9Tuesday

Bloomberg Technology

Musk’s xAI Taps Starlink Staffer to Run Grok Training Team

xAI brought in an executive from SpaceX’s Starlink service to run the Grok training team, replacing college-aged engineer Diego Pasini; the RSS snippet does not disclose the executive’s name, tenure, or training process.

Why it matters: xAI moved an executive over from SpaceX's Starlink division to run the Grok training team, replacing previous lead Diego Pasini.

Jun 6Saturday

AI HOT (Curated Pool)

SpaceX and Google Reach New Cloud Computing Agreement

SpaceX disclosed a cloud services agreement with Google: Google will pay SpaceX $920 million per month for computing capacity tied to xAI data centers, while the post does not disclose contract duration, GPU scale, or delivery terms.

Why it matters: SpaceX disclosed that Google pays $920 million a month for compute at xAI data centers, about $11 billion a year, making AI compute a basic resource that can be priced and contracted on its own.

Hacker News front page

Google to Pay SpaceX $920M a Month for Compute Capacity at xAI Data Centers

The title says Google will pay SpaceX $920 million per month for compute capacity at xAI data centers; the RSS snippet does not disclose contract duration, GPU scale, or the capacity delivery mechanism.

Why it matters: Google pays SpaceX $920 million a month for compute at xAI data centers, which works out to over $11 billion a year, a rare large-scale data center lease.

Jun 5Friday

Computing Life · Share · Yage

Grok Build 0.1: xAI’s Bet on Parallel Breadth

xAI launched Grok Build 0.1 in May 2026 as a coding agent built around parallel subagents; the post does not disclose benchmark results, cost figures, or specific privacy-policy terms.

Why it matters: xAI's coding agent Grok Build 0.1 runs 8 parallel sub-agents, a concrete contrast with Claude Code's single-agent depth-first approach.

Jun 4Thursday

AI HOT (Curated Pool)

OpenRouter compares 11 LLMs for real-time decisions: Claude and Grok lead

OpenRouter spent $482 on inference to run 11 LLMs through a 30-round real-time decision challenge, where Claude and Grok models led on decision speed and task success, while several high benchmark models underperformed on real-time scheduling.

Why it matters: OpenRouter spent $482 running 11 models through 30 rounds of a 2D battle royale, and the cost spread between winners is the point.

Financial Times · Technology

MP sues Musk’s xAI in UK test case over fake sexual images

UK MP Jess Asato sued Musk’s xAI over fake sexual images, using the claim to test whether AI model makers are liable for system outputs; the post does not disclose the model, generation mechanism, damages sought, or court timetable.

Why it matters: UK Labour MP Jess Asato is suing Musk's xAI over fake sexual images of her, a test case on whether model developers are liable for what their models output.

Jun 3Wednesday

AI HOT (Curated Pool)

Grok Becomes Vapi's Default Voice Engine

xAI partnered with Vapi to make Grok the default engine for 12 core voices, covering more than 2.5 million voice agents, and Grok Voice ranked first in Vapi’s independent blind test.

Why it matters: xAI and Vapi partnered to make Grok the default engine for 12 core voices on the platform, covering over 2.5 million voice agents, and Grok ranked first in Vapi's own blind tests.