Promptea Weekly
Week of 2026-10-11 to 2026-10-17
The big story
Gemini 4 Argon is generally available: $2/$10 until December, and the 1M-token limit is shared
Google opened its frontier model to API developers and paying subscribers on October 13. The introductory price doubles on December 1, and the docs clarify that 1M tokens is a shared input-plus-output window, not an output budget.
This week in AI
Gemini 4 Argon is generally available: $2/$10 until December, and the 1M-token limit is shared
Google opened its frontier model to API developers and paying subscribers on October 13. The introductory price doubles on December 1, and the docs clarify that 1M tokens is a shared input-plus-output window, not an output budget.
Google's frontier model is now generally available, with list pricing that undercuts Claude Opus 5.5 and GPT-6 Astra until November 30.
Read the full story →
Reflection releases Beam's weights: 501B parameters, Apache 2.0, and benchmarks it graded itself
The first model from the Nvidia-backed U.S. lab is now downloadable under a standard permissive license. Its efficiency pitch is promising but still unverified.
It is a frontier-scale U.S. open model under plain Apache 2.0, with no revenue clause or custom use terms, unlike Meta's Muse Spark Open released a day earlier.
Read the full story →
Meta releases Muse Spark Open: 816B open weights, a 1M-token window, and a license with a revenue clause
The open-weight Muse Spark is built for coding agents and can be self-hosted, but it needs hundreds of gigabytes of memory, and its capability claims are Meta's own.
It is a large, agent-oriented model that teams can run inside their own network, which matters for private codebases and for pinning model versions.
Read the full story →
Meta's first frontier model, Muse Wonder, reaches the API at frontier prices
Meta says Muse Wonder matches OpenAI and Anthropic's best on its own benchmarks. Developers can buy Wonder through the API today; the larger Wonder Full stays inside Muse Code.
A third vendor now sells a frontier-class model through an API at roughly the same price tier as OpenAI and Anthropic's top models, according to TechCrunch.
Read the full story →
NVIDIA RTX Spark laptops go on sale from $2,599, and the entry model has 24GB, not 128GB
Nine Arm-based Windows laptops from six PC makers are available today. The headline 128GB and 1-petaflop figures describe the top configurations.
Arm-based Windows laptops with CUDA are now a purchasable option for local model and agent work, not a roadmap item.
Read the full story →
NVIDIA pledges $250 million in compute for U.K. agentic AI safety research, with terms still to be agreed
The five-year donation is GPU access worth $50 million a year, routed through U.K. cloud partners. Oxford's BOLD, UCL's SOFAIR and Imperial get a share.
Agentic safety research is compute-hungry, and academic labs rarely have frontier-scale GPU access. A dedicated multi-year allocation changes what they can test.
Read the full story →