PUBLIC INTELLIGENCE
Selected AI market intelligence
Recent moves by company. Sign up for the full timeline, company profiles and trend analysis.
Model Studio / Bailian · major
Alibaba Cloud Bailian Token Plan Upgraded to Include Agent Tools
Alibaba Cloud Bailian's Token Plan subscription has been updated to shift its positioning from pure model token subscriptions to a 'Model + Agent Tools' bundle, emphasizing increased value without price hikes. This reflects the evolution of MaaS competition towards integrated subscription models, allowing developers and enterprises to access Agent tools directly within their subscription, thereby reducing the cost and barrier of building AI agent applications.
Model Studio / Bailian · major
Alibaba Bailian Introduces Qianwen Office AI Productivity Platform
Alibaba Cloud added 'Qianwen Office' to its AI application section, positioned as an all-in-one AI productivity platform and enterprise-grade agent product that delivers ready-to-use outcomes, marking a shift towards end-to-end agent application delivery.
Gemini · major
Gemini 部分音频/多模态模型移除上下文缓存定价
English summary is being prepared.
Gemini · major
Gemini 3.7 Flash 与 3.5 Flash 模型弃用及定价移除
English summary is being prepared.
Zhipu BigModel · major
智谱开放平台首页移除「模型微调」与「模型评测」入口
English summary is being prepared.
OpenAI Platform · major
OpenAI 新增生命科学专用模型系列 GPT-Rosalind
English summary is being prepared.
Volcano Engine Ark · major
Volcano Engine Updates Sample Code, Default Model Changed to doubao-seed-2-1-pro-260628
The default model parameter in all language sample codes (Python, Curl, Go, Java, OpenAI SDK) on the Volcano Engine product introduction page has been uniformly updated to 'doubao-seed-2-1-pro-260628'. The '260628' in the model ID suggests a version date of June 28, 2026, representing the latest iteration of the Doubao Seed 2.1 Pro series. As the platform's default recommended model, it will directly impact the initial experience and performance baseline for all new users, demonstrating ByteDance's rapid delivery and continuous iteration capabilities in foundational models.
Volcano Engine Ark · major
Volcano Engine Adds Cloud-Deployed MCP Capability Entry to Product Page
The 'Advanced Usage' section of the Volcano Engine product documentation has added a 'Cloud-Deployed MCP' entry, described as 'calling various vertical domain MCP tools to enhance model capabilities.' This indicates that the platform now officially supports the cloud deployment and invocation of MCP (Model Context Protocol) tools. As MCP becomes a standard for connecting AI platforms with external tools and data, this move enables Volcano Engine's tool ecosystem to interoperate with other MCP-supporting platforms, accelerating the standardization of the domestic Agent development ecosystem.
Anthropic · major
Anthropic 推出专注网络安全应用的「Anthropic 网络使命」计划
English summary is being prepared.
OpenAI Platform · major
OpenAI Adds Pricing and Ultrafast Low-Latency Mode for GPT-6.1 Sol
OpenAI has introduced two key updates for the GPT-6.1 Sol model. First, pricing has been added: short-context (≤272K tokens) input is $12.00/M tokens and output is $60.00/M tokens, with long-context prices roughly double, positioning it significantly cheaper than GPT-6 Astra. Second, an Ultrafast mode is now available in the Responses API, allowing developers to reduce output token latency by specifying `service_tier: "ultrafast"`. This mode is open to all API users and supports US/EU data residency, enriching OpenAI's model tier and inference speed options.
Anthropic · major
Anthropic Releases Cost-Effective Lightweight Model Claude Haiku 5.5
Anthropic released Claude Haiku 5.5, describing it as its fastest, cheapest, and most capable small model to date, designed for high-throughput, cost-sensitive workloads. This model further lowers the cost-performance baseline for lightweight models, directly pressuring the competitive landscape in high-throughput, low-cost inference scenarios (e.g., customer service, content moderation, data labeling) and lowering the barrier for batch and real-time applications.
AWS Bedrock · major
AWS Bedrock Adds Zhipu GLM 5.3 Model Support and Tiered Pricing
AWS Bedrock added On-Demand pricing for Zhipu's GLM 5.3 model, offering Standard, Priority, and Flex/Batch inference tiers across Global and US regions, expanding multilingual model choices for global developers.
AWS Bedrock · major
Amazon Bedrock Launches Public Preview of Managed Agents Powered by OpenAI
AWS announced the public preview of Amazon Bedrock Managed Agents powered by OpenAI, enabling users to build and deploy OpenAI-optimized agents within AWS's native governance and security framework.
Zhipu BigModel · major
Zhipu GLM-4.6 Model Removed from Hugging Face
Zhipu's GLM-4.6 text-generation model has been removed from the zai-org organization repository list on Hugging Face.
Microsoft Azure AI Foundry · major
Foundry Agent Service Introduces Managed Agent Network Egress Control
Microsoft Foundry now allows users to define and attach egress policies for managed agents, observe calls in audit, and validate allowed/denied destinations in enforce mode, enhancing enterprise security and compliance.
Microsoft Azure AI Foundry · major
Microsoft Foundry Routines GA: From Chatbots to Automated Agents
Microsoft Azure AI Foundry's Routines feature is now generally available (GA), supporting agents to monitor work, respond to business events, and continuously execute tasks, evolving from chat-based interactions to automated agents.
The above is a public selection. Sign up to access full timelines, per-source evidence and trend analysis.
Sign up for more intelligence →