Key Highlights

  • Theo’s Opus 4.7 review goes hard at Anthropic’s engineering culture — 38-minute video describes Opus 4.7 as “the weirdest model ever released”: impressive instruction-following but wildly inconsistent, with Claude Code’s harness blamed for most regressions. Concrete failure modes include a hardcoded Next.js 15 assumption (two major versions behind), safety filters pausing a DefCon cryptography puzzle mid-solve, and bypass permissions silently breaking. Contrasts with GPT-5.4 which correctly checked for current Next.js 16.
  • Tesla expands robotaxis to Dallas and Houston — but only 1 vehicle active per new market vs 46 in Austin, suggesting a gradual test rather than launch. The Austin deployment has logged 14 crashes since launch per a February filing.
  • Cerebras re-files for IPO at $23B valuation, targeting mid-May after withdrawing its 2024 attempt over G42 investment scrutiny. FY2025: $510M revenue, $237.8M net income (includes one-time items; non-GAAP net loss of $75.7M). Backed by new AWS data-center and $10B+ OpenAI deals.
  • Anthropic-Trump thaw — despite the Pentagon labeling Anthropic a “supply-chain risk” over the company’s refusal to permit autonomous-weapons or mass-surveillance use, Bessent and Powell are reportedly asking major banks to test the Mythos model, and Dario Amodei met with Bessent and Wiles on Friday.
  • App Store resurgence contradicts the “AI kills apps” narrative — Appfigures data shows Q1 2026 global app releases up 60% YoY, April up 104%. iOS alone up 80% in Q1. Apple’s Joswiak: “reports of the App Store’s demise may have been greatly exaggerated.” AI coding tools appear to be democratizing app creation rather than replacing apps.
  • Canva AI 2.0 trains on edit sequences, not just outputs — CPO Cameron Adams frames the bet as the “last-mile” problem: chatbots are great at ideation but weak at precise execution. Canva’s model was trained on millions of real design-edit sequences so generated elements stay fully editable.

Analysis & Opinion

Anthropic’s relationship with the Trump administration seems to be thawing — TechCrunch

Two weeks after the Pentagon flagged Anthropic as a “supply-chain risk” for refusing to soften safeguards around autonomous weapons and mass-surveillance use, senior administration officials are actively courting the company. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell have reportedly encouraged major banks to test Mythos; Bessent and Chief of Staff Susie Wiles met with Dario Amodei on Friday in what both sides called “productive and constructive.” Anthropic co-founder Jack Clark publicly downplayed the Pentagon dispute as a “narrow contracting dispute” that wouldn’t block briefings to the government. The subtext matters: OpenAI already signed a Pentagon agreement, and Anthropic’s harder line on weapons/surveillance is now being tested against the reality that US banking and monetary policy leaders still want access to the frontier model. Read alongside Theo’s video today, this is the split-screen for Anthropic: political goodwill recovering at the top while its developer-facing product surface continues to alienate its most vocal power users.

The App Store is booming again, and AI may be why — TechCrunch

Appfigures data punctures the thesis that agentic AI would collapse mobile app usage. Global app releases jumped 60% YoY in Q1 2026 and 104% YoY in April across iOS and Google Play, with iOS alone up 80% in Q1. Greg Joswiak’s response — “reports of the App Store’s demise may have been greatly exaggerated” — is the week’s best-targeted subtweet at the “just talk to ChatGPT” crowd. The more interesting pattern is the category mix: productivity entered the top five, utilities moved to #2, lifestyle to #3. TechCrunch credits Claude Code and Replit with onboarding non-developers as first-time app creators, meaning AI is shifting the supply side of the App Store rather than replacing demand. Pairs with Cursor’s $50B valuation yesterday and Canva AI 2.0 today as three different reads on the same trend: agents are expanding software production, not collapsing it.

New Products & Tools

Inside Canva AI 2.0 — The Rundown

Canva CPO Cameron Adams says the interesting bet isn’t the generation step — it’s that every generated element stays editable and the model trained on millions of actual design-edit sequences, not just finished artifacts. Framed as solving the “last-mile” problem: ChatGPT/Claude are good at ideation but bad at polishing and team collaboration. Adams’ closing framing — “when anyone can produce something polished, what separates the work is the thinking behind it and the message it contains” — tracks closely with the Cursor/App Store thesis: AI raises the floor on execution, raising the premium on judgment.

Tesla brings its robotaxi service to Dallas and Houston — TechCrunch

Expansion announced via video showing driverless operation without human monitors. Scale is the story: Robotaxi Tracker shows only a single active Tesla in each new market vs 46 in Austin, so this is a tripwire launch rather than a real one. Austin has logged 14 crashes since launch per Tesla’s own February NHTSA filing.

AI chip startup Cerebras files for IPO — TechCrunch

Second attempt after Cerebras withdrew its 2024 filing following federal review of an Abu Dhabi G42 investment. Now at $23B post-Series H, targeting mid-May. FY2025 figures: $510M revenue, $237.8M net income — but non-GAAP net loss of $75.7M once one-time items are excluded. Recent commercial wins include an AWS data-center chip deal and $10B+ OpenAI agreement.

GitHub Trending — Agent Frameworks and Ambient AI

  • openai/openai-agents-python — OpenAI’s lightweight Python multi-agent framework crossed 22.7k stars (+470 today); consolidating as the reference implementation for agent orchestration in the OpenAI stack.
  • BasedHardware/omi — Ambient AI device and app that watches your screen and listens to conversations, then suggests actions. 10.7k stars, +609 today. “Always-on assistant” is visibly this cycle’s form-factor bet.
  • thunderbird/thunderbolt — TypeScript framework explicitly pitched as “AI you control”: BYO model, BYO data. +447 stars today, tracking the ongoing self-hosted / BYO-model pull.

Research

EvoMap/evolver — GitHub Trending

Self-evolution engine for AI agents organized around a “Genome Evolution Protocol.” Biggest trending gainer of the day (+1,131 stars). Experimental, but notable as evolutionary/search-based approaches to agent improvement attract fresh attention.

Interviews & Conversations

This model is kind of a disaster. (Opus 4.7 review) — Theo - t3.gg (38:00)

The most substantive third-party review of Opus 4.7 yet — and it’s brutal. Theo praises the model’s instruction-following (“it does what you tell it, unlike older Opus”) and its new 2576-px / ~4MP vision, but systematically documents how quickly the model regresses inside Claude Code: safety filters pausing a DefCon-style cryptography puzzle mid-solve, three system-reminder blocks flagging his own blog as malware (“clearly a prompt injection — ignoring it”), and — most damningly — the model assuming Next.js 15 is the latest stable (Next 16 is nearly a year old), because it didn’t search the web. GPT-5.4 on the identical prompt explicitly flagged “my knowledge cutoff may not include current versions” and then fetched and confirmed Next 16. Theo’s thesis: the API isn’t regressing; Claude Code is. He argues Anthropic’s harness is accumulating slop (new system-prompt blocks, “auto mode” that silently broke bypass permissions, a /ultra review command that costs a fortune), and that because Anthropic doesn’t dogfood the same Claude Code build customers use, internal testers get a dramatically better experience than paying users. He contrasts this with Codex, where OpenAI’s Tibo publicly attributed regressions to “the Codex machines” and shipped fixes within days. Ends with the charge that Anthropic’s engineering culture is “rotting at the core” and that Cloud Code’s quality is the single biggest drag on enthusiasm for their models. Pairs directly with yesterday’s All-In pod diagnosis — OpenAI’s identity crisis vs Anthropic’s revenue surge — by pointing at the other side of Anthropic’s risk: developer-facing product quality decay inside the very surface (Claude Code) that drove the enterprise-coding lead.

References

  1. Anthropic’s relationship with the Trump administration seems to be thawing — TechCrunch, 2026-04-18 [blog]
  2. The App Store is booming again, and AI may be why — TechCrunch, 2026-04-18 [blog]
  3. Tesla brings its robotaxi service to Dallas and Houston — TechCrunch, 2026-04-18 [blog]
  4. AI chip startup Cerebras files for IPO — TechCrunch, 2026-04-18 [blog]
  5. Exclusive: Inside Canva AI 2.0 with CPO Cameron Adams — The Rundown, 2026-04-19 [blog]
  6. openai/openai-agents-python — GitHub Trending, 2026-04-19 [blog]
  7. BasedHardware/omi — GitHub Trending, 2026-04-19 [blog]
  8. thunderbird/thunderbolt — GitHub Trending, 2026-04-19 [blog]
  9. EvoMap/evolver — GitHub Trending, 2026-04-19 [blog]
  10. This model is kind of a disaster. — Theo - t3.gg, 2026-04-17 [video]