Strongly Supported Google has introduced the Fairwind Program, a limited access initiative providing governments and trusted partners with access to its cyber defense tools. The announcement was made alongside the introduction of Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, though details on these products are limited.
Strongly Supported Google plans to invest $1.5 billion in 2026 and 2027 to expand its data center campus in Jackson County, Alabama, which has been operating since 2019 on a repurposed site. The investment aims to strengthen the company's presence in the region and support community initiatives.
Provisionally Supported OpenAI's GPT-6 Astra model is now available on Unity Gateway, targeting enterprise use cases such as reasoning, document processing, coding, agentic search, multimodal inputs, and long-context workloads. Access is provided via Foundation Model APIs on a pay-per-token basis.
Provisionally Supported Databricks has announced that Google Gemini 3.8 Flash and Anthropic Claude Fable 5.1 are now available as Databricks-hosted models, accessible via Foundation Model APIs. Customers are reminded to ensure compliance with applicable terms.
Provisionally Supported Multiple incremental releases of llama.cpp (b10054 through b10064) include OpenCL transpose optimization for Q4_K, Vulkan support for Q2_0 quantization, SYCL fix for row calculation, and a test fix to avoid NaNs. Other changes include a default CPU routine for BLAS hadamard mul_mat and documentation updates.
Provisionally Supported Ollama released version 0.32.1 with improvements to Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations. The update also fixes a recurrent MLX model cache leak that could increase memory use across requests and improves cache snapshot performance.
Provisionally Supported The cline/cline project released version 0.0.61 of its SDK and version 3.0.41 of its CLI. Both updates include context compaction progress reporting and workspace git information persistence across sessions. The CLI also adds model ID suggestions from OpenAI-compatible endpoints and fixes a crash during compaction.
Provisionally Supported Several patches were merged into llama.cpp, including SYCL optimization for Battlemage GPUs (b9995), a new SME2 f32 kernel for ARM (b9999), Vulkan/CPU f16 SET_ROWS support (b10004), and bug fixes for DeepseekV4 seq_rm (b10005), export-graph-ops segfault (b10001), and log flushing (b9996). These changes enhance performance and stability across backends.
Provisionally Supported BerriAI released two new versions of litellm: v1.90.4 and v1.94.0-dev.1. Both releases include documentation on verifying Docker image signatures using cosign, referencing the same signing key from commit 0112e53.
Provisionally Supported Unity has published a series of articles detailing new AI-driven tools for game development, covering 2D and 3D asset generation as well as story creation. These tools aim to streamline the creative process for developers by automating certain aspects of content production.