Models & Labs
Google launches $899 'Googlebook' with baked-in Gemini AI
TechCrunch AIhigh confidence
Why it matters
- →Google is shifting from selling AI software subscriptions to anchoring them in dedicated hardware for the education sector.
- →The 'Googlebook' represents a strategic attempt to migrate 50 million Chromebook users into the Gemini ecosystem.
- →Bundling AI Pro with hardware lowers the barrier to entry for schools adopting generative AI tools.
Google has opened preorders for its new 'Googlebook' line, priced at $899, marking a significant shift in its Chromebook strategy by integrating Gemini AI directly into the operating system. The devices feature an AI-powered cursor and 'Rambler' dictation tool, designed to replace traditional point-and-click workflows with generative assistance. Targeting the K-12 sector, Google aims to transition millions of existing Chromebooks to this new platform, bundling 12 months of Google AI Pro with each unit. Flagship models from Acer, ASUS, Dell, HP, and Lenovo will launch in October across major global markets.
Read originalMore from TechCrunch AI
© TechCrunch AIGeneral AIresearch
OpenAI launches math advisory group amid controversy
OpenAI is establishing an independent advisory group at the Institute for Advanced Study to manage its expanding role in mathematical research. This move follows intense scrutiny after the company claimed its internal models solved over 100 open problems, including the Navier-Stokes Millennium Prize problem. The initiative aims to bridge the gap between AI labs and the mathematical community, addressing concerns raised by Fields Medalists about the pace and validity of AI-generated proofs. While the group can advise on significance and release coordination, it explicitly lacks authority to slow down OpenAI's internal progress. This signals a shift toward institutionalizing oversight in high-stakes scientific discovery without ceding strategic control.
TechCrunch AI
© TechCrunch AIMarket & Regulationother
Meta's Muse outpaces ChatGPT in early mobile downloads
Third-party data suggests Meta’s new AI app Muse is gaining traction faster than ChatGPT did at launch. In the U.S. and Canada, Muse hit 1.8 million iOS downloads in its first 12 days, beating ChatGPT’s 1.3 million. Daily active users also lead, with 642,000 in the U.S. compared to ChatGPT’s 231,000 at the same stage. This surge likely stems from Meta’s cross-promotion across Instagram and Facebook, where over 95% of Muse users are already present. The app has climbed to No. 1 on the U.S. App Store, signaling strong initial adoption for a consumer-facing AI tool.
TechCrunch AI
© TechCrunch AIInvestment · $1M
Market & Regulationbusiness
Tabby raises $1M to automate bookkeeping with AI
Tabby is betting that small businesses don't need better accounting software, they just need it gone. By using Plaid for live data and AI to generate real-time financial dashboards, the startup aims to replace the manual entry layer entirely. With 5,500 users and $100k ARR, they are proving there is demand for this frictionless approach against giants like QuickBooks. This pre-seed round signals that investors are backing tools that remove software complexity rather than just adding AI features to existing workflows.
TechCrunch AI
More in Models & Labs
Models & Labsmodels
vLLM v0.30.0: DeepSeek V4.1 and Fast Start
This release is a massive infrastructure overhaul for serving the latest reasoning models. The headline feature is native support for DeepSeek-V4.1-Flash, storing its entire KV cache in MXFP8 on SM100 hardware to drastically reduce memory overhead. For operators tired of slow cold starts, the new Fast Start daemon caches post-quantized weights in GPU memory, allowing engines to map over CUDA IPC instead of reloading from disk. It also brings HiSparse, a host-resident tier that spills KV pages to pinned host memory under pressure, effectively expanding usable context windows without buying more GPUs.
vLLM Releases
Models & Labsother
llama.cpp b11080 release with ROCm 10 and CUDA 13
This update quietly extends llama.cpp’s hardware support to ROCm 10.0 and CUDA 13.4 across Linux and Windows, keeping the library competitive as NVIDIA pushes newer driver stacks. The test suite also gains regex filtering for backend operations, a practical improvement for developers debugging specific inference paths. While no new model architectures are introduced, this release ensures compatibility with the latest GPU ecosystems without forcing users to wait for major version bumps.
llama.cpp Releases
Models & Labsother
llama.cpp b11081 release with CUDA 13 and ROCm 10 support
This release quietly cements llama.cpp as the universal inference runtime by adding default builds for CUDA 13.4 and ROCm 10.0, effectively closing the gap on newer NVIDIA and AMD hardware without requiring manual compilation flags. The inclusion of KleidiAI for Apple Silicon remains a key differentiator for local Mac users seeking optimized ARM kernels. While the changelog details internal test improvements like configurable tensor standard deviation, the real value lies in the expanded binary matrix that supports developers across the latest GPU architectures. Readers can now deploy on cutting-edge hardware out of the box. The update ensures that teams using the newest graphics cards do not need to wait for source code patches to achieve stable performance. This is a practical step toward making local inference truly hardware-agnostic.
llama.cpp Releases