AI Flash · live feed
A timeline of everything that has passed through the radar — newest first, with an evidence status for each item.
Continuously updated
Filter contains 225 items
How to read evidence statuses ↓
Granite-Switch-4.1-8B and Granite-Switch-4.1-30B have been released with adapters for decision-making tasks and integrated speech transcription. According to the authors, they require approximately the same GPU memory capacity as the base model alone.
The weights for Muse Spark Open from Meta are available on Hugging Face. The free license requires the attribution “Built with Muse Spark Open”; API providers with annual revenue above 20 million dollars need a separate agreement.
According to NVIDIA, the new NVIDIA DLSS 4.5 controls in GeForce NOW let members adjust the visuals and performance of supported games when playing in the cloud.
Part of a broader thematic overview; only this AI event has been covered.
Mistral AI has released Timestral, its first family of models for time series forecasting. The four variants have 5 to 209 million parameters and handle multiple related series as well as covariates. According to the company, they do not require a GPU to run.
The FLUX.3 Image Klein model from Black Forest Labs is available in the Diffusers library with two variants of the weights. It supports text-to-image generation, editing with multiple reference images, and control over the placement of elements using bounding boxes.
OpenAI announced native integration with Decisions in GPT-Live. According to the company, applications can trigger actions while the user is speaking or respond to the service's output, either through direct delegation or based on incremental transcript updates.
Liquid AI has made d1-ultra available through Liquid API. It returns answer probabilities based on text and images. The price is 0.042 USD per million input tokens; output is not billed, but each question counts as a separate input, including all images.
The Compliance API now returns the text of instructions that organization members have saved in claude.ai and Cowork. The feature is in beta for organizations with Claude Enterprise and requires the read:compliance_user_data permission.
GitHub Copilot offers an updated HydraFusion and a new HydraFusion Max variant for more demanding tasks. Both options remain in research preview and are available in VS Code, the GitHub Copilot app and GitHub Copilot CLI under the specified conditions.
LFM2.5-24B-A2B builds on LFM2-24B-A2B, expands the context from 32K to 131K tokens and always produces a reasoning trace before answering. According to the authors, it achieves 184 tokens/s on Apple M5 Max. The available excerpt does not confirm a public release.
Buttons for expressing interest or lack of interest may also appear below posts on Facebook. Meta may then ask follow-up questions that users answer in their own words. This is part of changes to recommendations in Feed and Reels.
GitHub has made AI Scan metrics available in public preview: finding counts, their resolution, and the use of suggested automatic fixes. The feature is available on github.com to customers with GitHub Code Security or GitHub Advanced Security; GitHub Enterprise Server is not supported.
Liquid AI has made LFM2.5-24B-A2B available for local agents. It has 24 billion parameters, with 2 billion active per token, and a context window of 131 072 tokens. According to the company, it generates 184 tokens per second on Apple M5 Max hardware.
Gemini 4 Argon became generally available in Gemini API on 13 October 2026 under the identifier gemini-4-argon. According to the provider, it is designed for complex programming, agentic tasks, mathematics and science.
According to GitHub, enterprise administrators can link GitHub Copilot activity to specific users without collecting prompts and responses. Five content categories can be controlled separately. Identity collection and collection of all content categories are disabled by default.
The GitHub Copilot plugin for JetBrains now requires JetBrains IDE 2025.2 or later. The update allows company administrators to set the default model, disable automatic startup of MCP servers, and open an inline chat with a request to fix a diagnostic issue.
The model gemini-3.5-flash is marked as deprecated in Gemini API. All requests to it are automatically redirected to the model gemini-3.6-flash. The announcement recommends updating the model identifier.
Ultrafast mode is available for the model GPT-6.1 Sol in the Responses API to all API users, subject to rate limits. According to OpenAI, it shortens the delays between generated output tokens and supports data residency in the USA and EU.
Compliance API now also includes chats from the unified Claude environment. This expansion is in beta for organizations using Claude Enterprise and uses the existing Compliance Access Key.
Compliance API now supports downloading documents from Claude Docs as Word files in beta. The announcement links to the Retrieve files and artifacts documentation.
Kevin Lin announced the availability of the open-source platform OpenClaw Enterprise for managing and deploying persistent AI agents at DevDay 2026. According to OpenAI, it adds customer isolation, policy management and security boundaries.
According to OpenAI, developers can define custom annotation controls in the ChatGPT app browser and programmatically trigger suggested prompts. Integration with WebMCP tools allows context from selected page elements to be passed to the agent.
Starting on the day of the announcement, GitHub is automatically moving existing AI credential checks to a new model. They remain available to GHSP and GHAS customers at no additional charge. According to the company, the model uses the context of the surrounding code to recognize passwords that lack a typical format.
GitHub has made local sandboxing generally available in GitHub Copilot CLI, the GitHub Copilot app and VS Code editor sessions with Agent Host. The feature, available at no additional charge, restricts access by running tools to system resources according to policies set by the developer or organization.
AI Radar monitors Czech and international sources every day, looking for changes that truly deserve attention.
The result is a quick overview of what has actually changed in the AI world, rather than another stream of articles.
Use the CS/EN switch to read the same Radar in Czech or English. English content is published after its translation has been checked, so new and older items may appear later.
Each row represents one event — not one article. At a glance, you can see its significance, credibility and main point.
Illustrative example, not a current news item.
Agent mode is available to all paying users
Until now, the mode was available only on the highest plan; it is now available on all paid tiers without a waitlist.
The detail page contains a fuller summary, its significance and original sources. Practical impact appears in the detail and the For individuals and For businesses views. An AI Flash item reaches the main selection only after it has been expanded and meets the publication rules.
We first summarize each event in the same way for everyone. Based on those same facts, we then explain what the change means for your own use and what it could mean for how a company operates.
Today’s briefing is the same for everyone. Pages For individuals and For businesses can be found in the main navigation — they select only events relevant to the given use case.
Reset history · Outage history · Subscribe in a feed reader (RSS) · Methodology and label explanations
