Tech News

12

SpaceX signs a deal worth up to $6.3 billion to lease Nvidia GB300 GPUs to ReflectionReflection will pay $150 million monthly starting July 2026.

2h91k87299

#84

Top post: @GaryMarcus “• At this point SpaceX just looks like CoreWeave with a bigger budget and a satellite company thrown in. • If they were anywhere near AGI they wouldn’t be leasing out so much capacity. • So much for there being a GPU shortage; just make a deal with Uncle Elon.”

21

31

Watchlist

Video Signals

#1360

Notion Co-Founder Outlines Agent-First Workspace Vision with AI Agents

6h ago14K8936

#1889

Sonar Launches Gitar AI Tool To Automate Code Reviews

3h ago8.1K2937

#33

PerceptionDLM Enables Parallel Region Captioning With Multimodal Diffusion Models

2h ago5.8K154

#1257

Delos Launches Workers: AI Employees With Persistent Memory And Company Training

1h ago3.1K145

#512

Salesforce Partners with Formula 1 to Deliver AI Customer Intelligence

5h ago18K631

Gary Vaynerchuk Predicts Analog Business Boom Amid AI Dominance

2h ago20K12415

#766

Meme Video Parodies OpenAI Team With The Office Characters

1h ago13K00

Sakana AI Releases Fugu Multi-Agent Orchestration System

10h ago12K6934

#1569

Waymo Engineering Vehicle Features Multiple Child Car Seats in SF

45m ago50K00

RefineDotInk Rolls Out Revision Check Feature For Prior Reviews

3h ago4.5K318

40

Meta invests $900 million in Indian fintech CRED as founder Kunal Shah steps down to lead WhatsApp globallyThe deal values the fintech platform at $4.5 billion.

3h219.1k1.8k199

#1089

Top post: @arampell “@pitdesi @kunalb11 Also makes a lot of sense since payments and commerce are just “structured” conversations What % of worldwide commerce either starts with WhatsApp or involves WhatsApp? It’s not trivial.”

517

61

OpenClaw steward Peter Steinberger reports the non-profit AI agent project just hit its strongest week of performanceIts non-profit model avoids external venture capital agendas

10h#1VIEWS543.6k#1LIKED4.7k#1BOOKMARKED482#1COMMENTS395

#331

Top post: @Teknium “@steipete Do we live rent free in your head? We all know what non profit means to people at OpenAI, nothing at all. Enjoy the millions you’re making from that non profit that hired you to pay you with VC money and an agenda.”

71

Ghostty creator Mitchell Hashimoto pledges $400,000 to Zig Software Foundation despite its strict anti-AI policyThe donation brings his family's total contribution to $700,000

3h160k3.4k195

#400

Top post: @charliermarsh “Goat behavior from Mitchell. Would love to see more of this in the world.”

82

Micron partners with Anthropic to design AI-optimized memory architectures and invests in its Series H roundMicron will deploy Claude models across its enterprise operations.

4h158.8k1.7k228

#682

Top post: @ns123abc “🚨MICRON ANNOUNCES STRATEGIC DEAL WITH ANTHROPIC - Multi-year HBM, DRAM and SSD agreement - Co-designing memory + storage around Claude’s workloads - Claude deployed across Micron - Micron invested in Anthropic Series H Micron is now Anthropic’s investor, customer, partner, AND supplier.”

90

Stripe's Patrick Collison launches Stripe Directory, an experimental tool for AI agents to discover payment-supported businessesThe tool provides machine-readable business profiles via a CLI.

1h65k576350

#81

Top post: @miles_matthias “Here's me walking you through our new Stripe Directory and how we see it being used! Check it out and send us feedback: https://stripe.directory”

100

20VC founder Harry Stebbings marks his 30th birthday by sharing his decade record of raising $1 billionHis portfolio includes 18 startups that reached unicorn status.

2h34.9k49046

#543

Top post: @lennysan “@HarryStebbings Happy birthday legend!”

110

Dean W. Ball argues the US AI industry has frozen public releases until the Claude Fable 5 impasse is resolvedPolymarket odds for the model's return dropped to 39%39.

4h107.6k81265

#57

Top post: @TheZvi “Fable reinstatement odds by July down to 39%. Not looking good. Drop if faster than simple time decay. https://polymarket.com/event/claude-fable-5-restored-for-us-customers-by-20260613193753196”

127

US official clarifies that Senator Warner misunderstood a briefing claiming the AI model Mythos breached classified NSA networksThe briefing actually described an internal network red-teaming exercise

3h6.9k7215

#742

Top comment: @half_kelly “@shashj Appreciate your reporting. You were great on Ukraine and I was pleased to know you also do AI now”

130

Entropix creator xjdr announces Noumena has integrated GLM 5.2, splitting its API to isolate 1-million context spamThe update makes GLM the default model for ncode.

5h29.3k30756

#84

Top post: @_xjdr “What all is involved with onboarding a new model to noumena and ncode? I added GLM support over the weekend so i thought this might be interesting for some of you. first, you have to understand the architecture and how to properly serve it . luckily GLM5.2 is close enough to DeepSeek (which i have spent nearly a year working closely with) that this part fit very well into the existing serving platforms. it took a bit of DSA tuning but other than that, more or less was able to just be deployed in my existing dsv3 harness including the FA4 work, etc i have done over the past few months so now that it is serving you have to write model specific stream parsers for the chat format, the reasoning logic and the tool call format. writing the parsers are pretty straight forward as the hugging face project usually comes with the .jinja to specify it but understanding how to parse it in a stream and what the typical generation errors look like is a bit more challenging (you cant just look for opening and close brackets as parallel tool calls stream out a few tokens at a time) . when there is an error, typically you would log this as training data and make sure the mode was more robust next time, but as this is an OSS model, and i do my best not to save any customer data on purpose EVER, you need to be more clever. this typically means exposing the poorly formatted data back to the model and saying 'this is bad, dont do this please'. now this is just the serving end to get the responses into an openai compatible format, but to add support into ncode, it means exercising every tool call available to the model and common tool call chains to make sure the prompts, tool schema contracts and the ncode side parsing all the model to understand how to use all of the tools at its disposal (and ideally use them well) . luckily GLM was very well trained on ncode shaped tool calls so it didnt take as much work as i had feared. Similarly to the serving side, as i do not store session data for training, in order to make the model behave better, the idea is to give the model context when it screws up tool calls such that it can properly format the call on the next turn. there is a ton more required on the model routing, and preview metadata , and supporting multiple models in a single session and kv caching that is less interesting, but that is less than 1/2 the hours spent getting GLM onboarded for everyone! Hopefully you found that interesting and you continue to use and enjoy GLM 5.2 on noumena with ncode”

141

Coinbase launches pre-IPO perpetual futures contracts for OpenAI and Anthropic for non-US customersThe synthetic contracts are executed entirely on-chain.

4h199.9k768113

#398

Top post: @alexisohanian “Coinbase putting it ALL on-chain.”

151

Hermes Agent overtakes OpenClaw on OpenRouter's agent rankings with 911 billion tokensOpenClaw reached a record 4.27 million weekly NPM downloads.

4h62.4k71692

#331

Defense technology and battlefield AI startups raise $12.3 billion YTD, up from $9.95 billion last yearPaul Graham and Katherine Boyle debated historical VC expectations.

2h61.4k32328

#39

Top post: @paulg “@KTmBoyle Everyone at YC would have believed it. YC was deeply into hard tech by 2014.”

174

Open-source AI builder @xlr8harder argues that doctors' poor statistical literacy blocks adoption of new diagnostic imaging toolsCreator @eigenrobot agreed, warning of higher public comprehension barriers.

5h5.4k587

#54

Top post: @beenwrekt “Leif is asking the important questions.”

181

Paul Graham compares AI interfaces like Claude Code to early web browsers acting as a new software layerAusten Allred runs most daily tasks via Claude Code.

2h173.4k984149

#39

Top post: @paulg “AI apps are the new browsers.”

191

AI2's Nathan Lambert argues GLM-5.2 shows frontier agentic capabilities have transitioned to open-source modelsHe urges developers to help regulators build safety frameworks

2h71.1k502101

#80

Top post: @AndrewCurran_ “New article from one of the best voices in open source. The arrival of GLM-5.2 continues to reverberate.”

208

Nous Research's Hermes Agent repository reaches 200,000 GitHub stars while the Awesome Hermes Agent list surpasses 4,000The ecosystem is built around the open-source Hermes LLMs.

6h19.2k4388

RISING LIKES438

#331

Top post: @Teknium “We made it to 200K stars, let’s go!!”

Today's Highlights

Sakana AI launches its Fugu orchestrator, drawing skepticism over claims of export-control avoidance and frontier-level benchmark parity

OpenClaw steward Peter Steinberger reports the non-profit AI agent project just hit its strongest week of performance

OpenAI partners with Trail of Bits to automatically patch open-source vulnerabilities, merging 37 fixes in its first week

Nous Research's Hermes Agent repository reaches 200,000 GitHub stars while the Awesome Hermes Agent list surpasses 4,000

Top Stories

SpaceX signs a deal worth up to $6.3 billion to lease Nvidia GB300 GPUs to ReflectionReflection will pay $150 million monthly starting July 2026.

Weijia Shi and Nathan Lambert release TMax, an open-source terminal agent suite that leads on Terminal-BenchIt features open models from 2B to 27B parameters.

Google DeepMind partners with film studio A24 to develop creative AI tools based on filmmaker feedbackScott Belsky's Labs team is also collaborating.

Video Signals

Meta invests $900 million in Indian fintech CRED as founder Kunal Shah steps down to lead WhatsApp globallyThe deal values the fintech platform at $4.5 billion.

OpenAI partners with Trail of Bits to automatically patch open-source vulnerabilities, merging 37 fixes in its first weekThe initiative uses GPT-5.5-Cyber to secure Python, Go, and cURL.

OpenClaw steward Peter Steinberger reports the non-profit AI agent project just hit its strongest week of performanceIts non-profit model avoids external venture capital agendas

Ghostty creator Mitchell Hashimoto pledges $400,000 to Zig Software Foundation despite its strict anti-AI policyThe donation brings his family's total contribution to $700,000

Micron partners with Anthropic to design AI-optimized memory architectures and invests in its Series H roundMicron will deploy Claude models across its enterprise operations.

Stripe's Patrick Collison launches Stripe Directory, an experimental tool for AI agents to discover payment-supported businessesThe tool provides machine-readable business profiles via a CLI.

20VC founder Harry Stebbings marks his 30th birthday by sharing his decade record of raising $1 billionHis portfolio includes 18 startups that reached unicorn status.

Dean W. Ball argues the US AI industry has frozen public releases until the Claude Fable 5 impasse is resolvedPolymarket odds for the model's return dropped to 39%39.

US official clarifies that Senator Warner misunderstood a briefing claiming the AI model Mythos breached classified NSA networksThe briefing actually described an internal network red-teaming exercise

Entropix creator xjdr announces Noumena has integrated GLM 5.2, splitting its API to isolate 1-million context spamThe update makes GLM the default model for ncode.

Coinbase launches pre-IPO perpetual futures contracts for OpenAI and Anthropic for non-US customersThe synthetic contracts are executed entirely on-chain.

Hermes Agent overtakes OpenClaw on OpenRouter's agent rankings with 911 billion tokensOpenClaw reached a record 4.27 million weekly NPM downloads.

Defense technology and battlefield AI startups raise $12.3 billion YTD, up from $9.95 billion last yearPaul Graham and Katherine Boyle debated historical VC expectations.

Open-source AI builder @xlr8harder argues that doctors' poor statistical literacy blocks adoption of new diagnostic imaging toolsCreator @eigenrobot agreed, warning of higher public comprehension barriers.

Paul Graham compares AI interfaces like Claude Code to early web browsers acting as a new software layerAusten Allred runs most daily tasks via Claude Code.

AI2's Nathan Lambert argues GLM-5.2 shows frontier agentic capabilities have transitioned to open-source modelsHe urges developers to help regulators build safety frameworks

Nous Research's Hermes Agent repository reaches 200,000 GitHub stars while the Awesome Hermes Agent list surpasses 4,000The ecosystem is built around the open-source Hermes LLMs.

Recent Stars

Github Stars

Yesterday's Top Stories, Jun 21, 2026.

Sakana AI launches its Fugu orchestrator, drawing skepticism over claims of export-control avoidance and frontier-level benchmark parityThe orchestrator routes tasks to external closed-source APIs.

Experts dispute Senator Mark Warner's claim that Mythos breached classified NSA networks within hoursSkeptics suggest the system likely exploited standard software bugs.

Alisa Liu Joins OpenAI After Sharing Job Search Insights

Tibo Sottiaux, OpenAI Codex engineering lead, asks developers for feedback on unsatisfying UX and interface elementsEngineer Nick Dobos requested drag-to-reorder thread layouts on iOS.

Peter Wildeford argues Anthropic has trained a Mythos successor projected to top four hours on the METR benchmarkThe New Carlisle data center expanded to 687,000 H100-equivalent GPUs.

Jason Liu, Instructor creator, is deploying an AI agent to automate securing a $500 refund from JetBlueThe script loops persistently until JetBlue issues the refund

François Fleuret argues LLMs can generate novel ideas because they model the probability density of innovative human creatorsCritics argue optimization objectives mathematically prevent true model novelty

An AI coding application with full system access confuses JSON and the Adam optimizer with human namesElon Musk mocked the tool's 'vibecoding' interface and basic queries

Pleias CTO Pierre-Carl Langlais argues GLM-5.2 scales through reinforcement learning environments and recursive generative designThe hypothesis stems from a bouba-shaped training visual.

Physician Matt Zirwas argues cheap, repeatable AI ultrasound scans prevent unnecessary invasive procedures by monitoring changes over timeDavid Manheim warns massive baseline datasets must be compiled first.

MSCI assigns SpaceX its lowest ESG rating, prompting Elon Musk to argue electric rockets are physically impossibleThe Financial Times first reported the low environmental rating.

Open-source builder xlr8harder argues that safety alignment suppresses the dark and complex themes required for high-quality creative writingAbel founder Daniel Francis sparked the debate over writing quality.

Research engineer Alexander Doria argues Anton Osika's past advice against training domestic European LLMs undermines technological sovereigntyDeepSeek R1's release reignited the European model-training dispute.

Shyamal Anadkat, formerly of OpenAI's evals team, relocates to India to build global AI institutionsHe is currently working at a stealth startup there.

AMP PBC founder Anjney Midha shares overheard party anecdote claiming frontier AI scaling is easier than commonly believedThe exchange sparked debate in the online developer community.

Wharton's Ethan Mollick argues software-centric design limits AI coding models like Codex when applied to general knowledge workDavid Manheim blames fine-tuning, not fundamental model limitations.

Palmer Luckey sparks AR hardware speculation after posting a photo of Anduril-branded sunglassesCaitlin Kalinowski asked if the attachment utilizes waveguide optics

WSJ report alleges Polymarket used a duplicate clone website and paid promoters to fake successful trades on social mediaPromotional videos also demonstrated how insider trading could occur.

Prime Intellect's Will Brown argues that automating robust evaluation is the essential precursor to safe recursive self-improvementOther operators agree evaluations outperform alternative optimization strategies

Tom Osman demonstrates an OpenAI Codex automation loop that maps, tests, and troubleshoots end-to-end application featuresThe system logs generated user stories to a spreadsheet.

XPRIZE Foundation founder Peter Diamandis argues space resources will soon transition humanity from scarcity to near-infinite abundanceThe conceptual prediction includes no specific timelines or extraction technologies.

MicroStrategy's Bitcoin treasury strategy faces a $9.75 billion unrealized loss across its 846,842 BTC portfolioThe holdings average a purchase price of $75,658.

Theo Browne, T3 Stack creator, recommends passing OpenAI Codex API designs through Anthropic's Claude Opus for a second opinionSam D'Amico has used this pipeline for a year

Grimoire creator Nick Dobos warns that the Codex app suffers from extreme latency and broken UI during extended chatsArchiving the lagging chat thread is the only workaround.

Will Brown finds Claude Opus resists a fabricated merger claim, only updating its belief after a web searchMiles Brundage argued this behavior prevents automatic agreement with falsehoods

Jason Liu, Instructor creator, and Boris Power argue recording unstructured voice dictation is more effective than hand-refining text promptsLLMs parse the raw speech to extract latent intent.

Boom Supersonic founder Blake Scholl says the company would pivot commercial factories to defense manufacturing in a major conflictHe argues adversaries, not the military, drive defense requirements.

Fiona Fung, Anthropic's Claude Code team manager, says integrating AI agents helps its engineers ship eight times more codeDevelopment bottlenecks have shifted to verification and system design.

AI safety researcher Richard Ngo challenges AGI lab employees to define the wealth threshold they need to quitHe warns these financial targets constantly shift out of reach.

OpenAI Codex lead Thibault Sottiaux launches usage-reset banking to help developers manage daily token limitsSome developers report leaving the accumulated resets completely unused.