OpenAI's Astra Lands With a Thud as Google Grabs the Spotlight Before Anthropic's IPO
OpenAI timed GPT-6.2 Astra to undercut Anthropic's IPO. The reviews came back brutal, and Google — not Anthropic — came out of the shadows to take the No. 2 spot in the AI race. Nvidia still posted a record $115.8 billion quarter, while export curbs against China threaten 20-30% of its revenue.
- Original source: finance.yahoo.com.
- The source published it on Nov 20, 2026 at 12:00 PM.
- No second outlet has reported it yet. Treat it as a single-source claim.
OpenAI rolled out GPT-6.2 Astra as a kill shot aimed right before rival Anthropic's IPO. It backfired. Yahoo Finance host Ejaaz Ahamadeen, who spent days testing the model, said early reviews were none other than absolutely appalling, with testers calling it an outright flop.
Ahamadeen's own take: Astra feels unfinished. It rambles, buries useful answers in complexity, and spirals into talking to itself instead of finishing the job. Side by side with its predecessor, GPT 6.1, the newer model looks worse — benchmarks barely moved, coding work falls short of Anthropic's Fable 5.2, and long-horizon tasks often go unfinished. Reviewer Theo publicly tore into it, saying the model's feel — how it understands you — simply is not there. Astra does shine at visual generation when paired with tools like Blender, and it reasons well when prompted very carefully.
The bigger surprise is who capitalized. Google shipped a run of new Gemini 4 models in recent weeks, and its latest beats GPT-6.2 Astra while competing with and beating Anthropic's Opus 5.5 and Fable 5.2 — at a lower price, according to Ahamadeen. He says the race is now down to Google and Anthropic, both chasing recursive self-improvement and disease research they have poured billions into.
Nvidia, meanwhile, posted a record $115.8 billion quarter without Jensen Huang, Ahamadeen noted. He also flagged that chip export restrictions aimed at China put roughly 20-30% of Nvidia's revenue in danger.
Anthropic went public the day before the video published, and Ahamadeen questioned why OpenAI would rush a model out ahead of a competitor's listing and risk its own frontier-lab reputation for a blow that never landed. He traced OpenAI's compute bets to math and cybersecurity, pointing to an unreleased internal model that the company says is 2 to 3x better at cybersecurity exploits than any other out there, including Anthropic's most powerful.
TopicsOpenAI · GPT-6.2 Astra · Anthropic · Google · Gemini 4 · Nvidia · Sam Altman · Ejaaz Ahamadeen
Source: finance.yahoo.com
Written by AI from verified reporting.
More stories
- Tech & AI semafor.com 0 viewsOpenAI Starts Vetting the Middlemen That Resell Access to Its Models
- Tech & AI coinpedia.org 1 viewAI chip rout broadens: Nvidia, AMD, Micron slide as Nasdaq sinks 2.5%
- Tech & AI tokenpost.com 1 viewAnthropic's IPO reignites AI trade as gold and Bitcoin hedge debt
- Tech & AI cryptopolitan.com 2 viewsTrump Says China Can Buy Only Older US Chips as Chip Stocks Sink

