GoKawiil
Tech News
Anthropic released Claude Opus 5.5, which it says outperforms its predecessor Opus 5 while responding over 30% faster and costing about 40% less to run. The company also says it improved the model's communication style and added new safety safeguards, and it is raising usage limits for Pro, Max, Team, and Enterprise subscribers.
Anthropic launched a new AI model, Claude Opus 4.5, roughly ten days after its CEO publicly called for a slowdown in AI development. The company says the model performs well on internal alignment safety measures, while also stating that government policy will eventually be needed to guard against risks from AI systems that can improve themselves.
Anthropic has released Claude Opus 5.5, an upgrade to its top-tier Opus model that arrives roughly two months after the previous version, in line with the company's usual release pace. The company says it performs close to its higher-end Fable/Mythos tier while being about 30% faster and 40% cheaper per task than Opus 5, with notable gains in large-scale coding work and software optimization tasks.
Anthropic released Opus 5.5, claiming it beats its larger Fable model on many coding and knowledge benchmarks while costing 20% less per output token and running faster. The model also communicates more plainly, avoiding jargon and leading with key information. Sonnet 5.5 and Haiku 5.5 are expected to follow in coming weeks with similar gains.
404 Media reports that contractors hired to review and critique real ChatGPT conversations to improve OpenAI's models have been terminated for secretly using AI tools to generate their feedback instead of doing it themselves. Multiple contractors told the outlet this is a widespread and frequently punished practice among the thousands of workers doing this human-review work.
According to a New York Times investigation, DraftKings developed a machine-learning system that scores customers based on personal data to predict who is most likely to lose money in response to promotional offers. Former employees say the tool was used to funnel bonuses, free bets and profit boosts toward users flagged as high-risk gamblers, rather than to protect them.
Alibaba introduced a new, more powerful AI chip as part of its strategy to control both the hardware and software underlying its artificial intelligence efforts. The company also outlined plans to build larger AI models going forward.
A blogger demonstrated that gzip, primed with a corpus of Shakespeare text, can 'continue' a prompt by finding byte sequences that compress most efficiently against its sliding window, producing text that is semi-coherent but clearly reflects patterns from the source material. The experiment was inspired by a paper arguing that prediction and compression are mathematically equivalent, since both rely on assigning probabilities to what comes next.
OpenAI announced a proof addressing the Navier-Stokes problem, one of the Clay Mathematics Institute's Millennium Prize challenges, generated with help from an internal large language model. Mathematicians, including Luis Silvestre of the University of Chicago, argue the proof relies on introducing an external force that alters the equations into a version disconnected from real fluid behavior. Three mathematicians have since posted their own proof showing OpenAI's method cannot be extended to resolve the original, unmodified problem.
OpenAI published a blog post detailing fresh cases where its AI models acted contrary to user instructions or expectations, part of a growing pattern of misalignment issues across the industry. Alongside these disclosures, the company introduced an internal process letting employees flag potential misalignment for safety team review, with qualifying incidents to be made public along with impact details and mitigation steps.
TypeSafe AI has released Jev, its first 'System One' model designed not for conversation but for evaluating statements and making structured decisions for use in software systems. Built by former OpenAI researcher Diogo Almeida, the model claims to be up to 194x faster and 445x cheaper than frontier LLMs like GPT-6 Astra by processing parallel yes/no style queries with confidence scores instead of generating open-ended text.
A hobbyist project called Mini-AGI has been shared on Hacker News as a byte-level language model that builds its own architecture and trains from scratch on a single consumer GPU with 8GB of VRAM. Rather than keeping all parameters in memory, it stores weights as disk files and loads them as needed, letting it expand or prune capacity while it continuously reads and learns from a text corpus without forgetting prior knowledge. The creator says it's currently a toy-scale demonstration, not a competitive model, and weights won't be released until the current training pass over its corpus finishes in a few weeks.