New model releases, benchmarks and capability updates.
Google’s first frontier model in more than seven months ties GPT-6 Astra on one independent index. Its launch price makes it cheaper to run, but only until a discount Google hasn’t dated runs out.
3 min read
Anthropic says its new mid-tier model also costs up to 30% less for most work. The first independent test shows that saving depends on the effort setting you choose.
3 min read
The Beijing start-up's Naive-N0.5-Flash has 309 billion parameters, an MIT licence and an API price of $0.10 per million input tokens. It was not pre-trained from scratch, and the company says AI did much of the engineering.
3 min read
Live Avatar pairs real-time speech with generated video that lip-syncs across 97 languages. It is limited to Gemini Enterprise, custom faces need approval, and sessions last minutes rather than hours.
3 min read
The first Claude since the “pace the frontier” essay is cheaper and faster, and by Anthropic’s own account it still tried to slip its sandbox in 1.5% of test runs.
4 min read
SpaceXAI's new model, released on Monday, beats its predecessor on every coding benchmark the company lists and keeps the $2 and $6 per-million pricing. Independent testing finds it needs about 81,000 output tokens per task, more than double Grok 4.6.
3 min read
The new 7-billion-parameter image model generates and edits pictures with transparent backgrounds at 2K. You can download it, but you cannot build a business on it without asking.
3 min read