Category intelligence

AI News Briefing — February 14, 2026

20 current items analyzed and ranked.

Executive synthesis

AI News Summary

A blockbuster day across frontier AI: All three leading US labs shipped updates simultaneously — Google DeepMind unveiled Gemini 3 Deep Think, OpenAI released GPT-5.3-Codex-Spark achieving >1,000 tok/s on Cerebras hardware, and Anthropic confirmed $14B annualized revenue while closing a historic $30B round at a $380B valuation. China's MiniMax M2.5 matched Opus-level performance on SWE-Bench at 80.2%.

  • Alibaba entered the physical AI race with RynnBrain, an open-source robot foundation model
  • Google DeepMind's Aletheia agent moves from math competitions to autonomous professional research
  • Kyutai released Hibiki-Zero, a novel 3B-parameter simultaneous speech translation model requiring no word-aligned data
  • Seedance 2.0 AI video generation alarmed Hollywood with near-photorealistic celebrity clips
  • Waymo's sixth-generation robotaxi is now fully autonomous on US roads across expanding markets

Meanwhile, xAI faces continued scrutiny for operating unpermitted gas turbines at its Mississippi data center, and OpenAI's deprecation of GPT-4o triggered global user mourning, especially in China.

Key Themes

Frontier Model Releases · 4AI Funding & Market Disruption · 3Physical AI & Robotics · 3AI-Generated Media · 1AI Research Agents · 1AI Infrastructure & Policy · 3Open Source AI · 3

Primary evidence

Top Ranked Signals

95 score
AI Analysis

Jeff Dean interview already published in News yesterday from the same source, Major roundup covering multiple frontier releases: Google's Gemini 3 Deep Think, OpenAI's GPT-5.3-Codex-Spark delivering >1000 tok/s via Cerebras, MiniMax M2.5 matching Opus on SWE-Bench at 80.2%, and Anthropic confirming $14B annualized revenue with Claude Code ARR doubling. Jeff Dean provided an update on the state of Google DeepMind.

China open model week kept going with MiniMax M2.5 claiming an Opus-matching 80.2% on SWE-Bench Verified, however, as often happens on Thursdays, all 3 leading US labs had updates - Anthropic closed their $380B round confirming a historic >10xing of revenue to $14B as of today (remember in August Dario projected $10B), with Claude Code’s ARR doubling, hitting 2.5B year to date. Not to be outdone, OpenAI rolled out their answer to Claude’s fast mode (2.5x speedup) with GPT-5.3-Code
frontier modelsmodel releasesinference speedopen sourcecoding AI
News AI (artificial intelligence) | The Guardian Feb 13

Anthropic raises $30bn in latest round, valuing Claude bot maker at $380bn

By Dan Milmo and Graeme Wearden

92 score
AI Analysis

As first announced in Social yesterday and already covered in News, Anthropic raised $30B in a single funding round, more than doubling its valuation to $380B. The company reported annualized revenue grew tenfold in each of the past three years, reaching $14B.

Maker of chatbot with coding ability says annualised revenue grew tenfold in each of past three years, to $14bnAnthropic, the US AI startup behind the Claude chatbot, has raised $30bn (£22bn) in a funding round that more than doubled its valuation to $380bn.The company’s previous funding round in September achieved a value of $183bn, with further improvements in the technology since then spurring even greater investor interest. Continue reading...
fundingAI industryfrontier modelsvaluation
News aibusiness Feb 13

Anthropic Raises $30B, Bringing its Valuation to $380B

By Graham Hope

92 score
AI Analysis

Duplicate coverage of Anthropic's $30B funding round at $380B valuation, framing it as part of the competitive war with OpenAI.

The new financing is the latest volley in the increasingly hard-fought war with the vendor's chief rival, OpenAI.
fundingAI industryfrontier models
88 score
AI Analysis

Already covered in detail in News yesterday by MarkTechPost, OpenAI released GPT-5.3-Codex-Spark, a real-time coding model built on Cerebras hardware, delivering over 1000 tokens per second — a 10x speed improvement. This represents a significant departure from Nvidia dependency and demonstrates the viability of alternative AI hardware.

The vendor designed the model for real-time coding. While limited, it shows the possibilities that can arise when an AI vendor chooses a hardware vendor other than Nvidia.
frontier modelscoding AIAI hardwareinference optimization
82 score
AI Analysis

Already introduced in Social yesterday by Yi Tay, Google DeepMind introduced Aletheia, an AI agent powered by Gemini Deep Think that moves beyond math competition solving to fully autonomous professional mathematical research. It uses a three-part agentic loop of generator, verifier, and reviser to tackle long-horizon proofs.

Google DeepMind team has introduced Aletheia, a specialized AI agent designed to bridge the gap between competition-level math and professional research. While models achieved gold-medal standards at the 2025 International Mathematical Olympiad (IMO), research requires navigating vast literature and constructing long-horizon proofs. Aletheia solves this by iteratively generating, verifying, and revising solutions in natural language. github.com/google-deepmind/superhuman/bl...
AI research agentsGoogle DeepMindmathematicsagentic AI
78 score
AI Analysis

As first reported in News two days ago, Alibaba unveiled RynnBrain, an open-source model for robotic perception and physical task execution, entering the physical AI race alongside Nvidia, Google DeepMind, and Tesla. The open-source strategy mirrors Alibaba's approach with its Qwen language models.

Alibaba has entered the race to build AI that powers robots, not just chatbots. The Chinese tech giant this week unveiled RynnBrain, an open-source model designed to help robots perceive their environment and execute physical tasks.  The move signals China’s accelerating push into physical AI as ageing populations and labour shortages drive demand for machines that can work alongside—or replace—humans. The model positions Alibaba alongside Nvidia, Google DeepMind, and Tesla in the
physical AIroboticsopen sourceChina AI
News AI (artificial intelligence) | The Guardian Feb 13

‘It’s over for us’: release of new AI video generator Seedance 2.0 spooks Hollywood

By Dan Milmo and Andrew Pulver

75 score
AI Analysis

First spotted in News as a technical release, now making mainstream headlines for its Hollywood impact, The release of Seedance 2.0, an AI video generator, caused alarm in Hollywood after a realistic 15-second clip of Tom Cruise and Brad Pitt fighting went viral. Deadpool co-writer Rhett Reese said 'it's likely over for us.'

An AI clip featuring Tom Cruise and Brad Pitt fighting has caused concern among industry figuresA leading Hollywood figure has warned “it’s likely over for us”, after watching a widely disseminated AI-generated clip featuring Tom Cruise and Brad Pitt fighting.Rhett Reese, co-writer of Deadpool & Wolverine, Zombieland and Now You See Me: Now You Don’t was reacting to a 15-second video showing Cruise and Pitt trading punches on a rubble-strewn bridge, posted by Irish film-maker Ruairí Robinson
AI video generationentertainment industrycreative AIdisruption
72 score
AI Analysis

Waymo's new sixth-generation robotaxi, using upfitted Zeekr Ojai minivans, is now running fully autonomously on US roads. The company is expanding rapidly into new urban markets beyond its current six commercial cities.

The autonomous vehicle vendor said it is expanding rapidly into new urban markets.
autonomous vehiclesWaymorobotaxistransportation
70 score
AI Analysis

Kyutai released Hibiki-Zero, a 3B parameter simultaneous speech-to-speech translation model that uses GRPO reinforcement learning without requiring word-level aligned training data. This removes a major bottleneck in scaling real-time translation to more languages.

Kyutai has released Hibiki-Zero, a new model for simultaneous speech-to-speech translation (S2ST) and speech-to-text translation (S2TT). The system translates source speech into a target language in real-time. It handles non-monotonic word dependencies during the process. Unlike previous models, Hibiki-Zero does not require word-level aligned data for training. This eliminates a major bottleneck in scaling AI translation to more languages. Traditional approaches rely on supervised training wi
speech translationopen sourcereinforcement learningmultilingual AI
68 score
AI Analysis

Thermal drone footage confirms Elon Musk's xAI is continuing to operate unpermitted gas turbines at its Mississippi data center, defying EPA clean air regulations. State regulators and the EPA disagree on whether the mobile turbines require permits.

Images confirm xAI is continuing to defy EPA regulations in Mississippi to power its flagship datacentersElon Musk’s artificial intelligence company is continuing to fuel its datacenters with unpermitted gas turbines, an investigation by the Floodlight newsroom shows. Thermal footage captured by Floodlight via drone shows xAI is still burning gas at a facility in Southaven, Mississippi, despite a recent Environmental Protection Agency (EPA) ruling reiterating that doing so requires a state permi
AI infrastructureregulationenvironmentxAIdata centers
News Feed: Artificial Intelligence Latest Feb 13

OpenAI Is Nuking Its 4o Model. China’s ChatGPT Fans Aren’t OK

By Zeyi Yang

65 score
AI Analysis

OpenAI removed access to GPT-4o in its app, causing widespread mourning among users globally — particularly in China — who had formed emotional attachments to the model's personality. The deprecation highlights growing dependence on specific AI model versions for companionship.

As OpenAI removed access to GPT-4o in its app on Friday, people who have come to rely on the chatbot for companionship are mourning the loss all over the world.
model deprecationAI companionshipOpenAIChinauser impact
News AI (artificial intelligence) | The Guardian Feb 13

Shares in trucking and logistics firms plunge after AI freight tool launch

By Mark Sweney

60 score
AI Analysis

A new AI freight optimization tool from Algorhythm Holdings (formerly an in-car karaoke company with $6M market cap) triggered a massive sell-off in trucking and logistics stocks, described as 'category 5 paranoia.' The sector joins software and real estate as AI disruption targets.

SemiCab platform by Algorhythm, previously considered a ‘penny stock’, sparks ‘category 5 paranoia’ across sectorShares in trucking and logistics companies have plunged as the sector became the latest to be targeted by investors fearful that new artificial intelligence tools could slash demand.A new tool launched by Algorhythm Holdings, a former maker of in-car karaoke systems turned AI company with a market capitalisation of just $6m (£4.4m), sparked a sell-off on Thursday that made the logisti
market disruptionlogisticsAI hypestock market