Meta launched Muse Spark 1.1 with a 1M token context window and context compaction, taking direct aim at OpenAI and Anthropic in the multi-agent AI race.
Meta’s Muse Spark 1.1 Puts a Million-Token Bet on Agentic AI Read More »
Chinese AI models now capture up to 46% of U.S. enterprise token traffic on OpenRouter, running 60-90% cheaper than American frontier systems as the cost gap reshapes the inference stack.
Chinese AI Models Capture 46% of U.S. Enterprise Tokens as Cost Gap Widens Read More »
Microsoft replaces OpenAI and Anthropic in Excel, Outlook, and 365 Copilot with in-house MAI models, cutting costs up to 70x and ending years of AI dependency.
Microsoft is throwing $2.5 billion and 6,000 engineers at a single problem: helping its biggest customers actually use artificial intelligence. The new unit, internally branded Frontier Company, marks the most aggressive enterprise AI deployment push from a hyperscaler to date and signals that the next phase of the AI race is not about who has
Microsoft’s $2.5B Frontier Bet: Why the AI Race Just Shifted From Models to Deployment Read More »
Microsoft unveiled a $2.5 billion initiative on Thursday that will place 6,000 of its own engineers, architects, and project managers inside customer companies to help them ship generative AI products. The new unit, branded the Frontier Company, marks the software giant’s most ambitious enterprise services bet since the launch of its OpenAI partnership and signals
Anthropic launches Claude Sonnet 5 on July 1, 2026, closing the gap to Opus on reasoning benchmarks. Day-one availability across AWS, GCP, Snowflake.
Anthropic Launches Claude Sonnet 5: Closing the Gap to Opus Read More »
Trump administration lifts export restrictions on Anthropic’s Claude Mythos 5 and Fable 5 AI models after months of regulatory friction with enterprise customers.
Anthropic Wins Lifting of Export Restrictions on Claude Mythos 5 and Fable 5 Read More »
Google delays the Gemini 3.5 Pro launch to July as the DeepMind team refines tool-calling, alignment, and red-team coverage. Here is what is being tweaked and why.
Google Delays Gemini 3.5 Pro Launch to July as It Tunes the Frontier Model Read More »
Reuters reports that enterprises are switching from premium to cheaper AI models as inference bills soar. Here is how the cost-conscious procurement trend is reshaping the AI vendor landscape and the architecture of corporate AI stacks.
Microsoft has thrown down a serious challenge to the artificial intelligence establishment with the release of its MAI family of models, a suite of seven purpose-built language and reasoning systems that scored 97 percent on the American Invitational Mathematics Examination benchmark. The announcement, published late Friday through the company’s internal research channels, signals that the
Microsoft Unveils MAI Family of Seven AI Models, Scoring 97% on Elite Math Benchmark Read More »
