Reports
AI-generated structured vendor updates
OpenAI Releases GPT-5.5 Instant with 52.5% Hallucination Reduction as New ChatGPT Default
<p>OpenAI released GPT-5.5 Instant, replacing GPT-5.3 Instant as the default ChatGPT model. Hallucination rate in high-risk domains dropped 52.5%, AIME 2025 math score 81.2 (vs 65.4 prior), GPQA 85.6 (vs 78.5). Response length reduced 30.2%. New "memory sources" feature lets users see which conversations/files/Gmail the model referenced. First Instant model flagged as High Capability (cybersecurity/biochemical domains). Available via chat-latest API.</p>
OpenAI Releases GPT-5.5 Instant with 52.5% Hallucination Reduction as New ChatGPT Default
<p>OpenAI released GPT-5.5 Instant, replacing GPT-5.3 Instant as the default ChatGPT model. Hallucination rate in high-risk domains dropped 52.5%, AIME 2025 math score 81.2 (vs 65.4 prior), GPQA 85.6 (vs 78.5). Response length reduced 30.2%. New "memory sources" feature lets users see which conversations/files/Gmail the model referenced. First Instant model flagged as High Capability (cybersecurity/biochemical domains). Available via chat-latest API.</p>
OpenAI-Microsoft Restructure: End of Exclusive AI-Cloud Era
This deal's end is an inevitable result of Anthropic's competitive pressure. What OpenAI lost is not just Azure's exclusive distribution but also the enterprise trust endorsement from the 'Microsoft ecosystem'. For the industry, the matrix of three major model vendors (OpenAI, Anthropic, Google) + three cloud vendors (AWS, Azure, GCP) is forming, shifting competition from '渠道为王' to 'model capability as king'.
Cerebras Launches IPO with $20B OpenAI Deal
AI chipmaker Cerebras filed for US IPO on Nasdaq with ticker CBRS; secured $20B multi-year deal with OpenAI to deploy 750MW of chips.
OpenAI Closes $122B Largest Private Funding Round
OpenAI closed the largest private funding round of $122 billion, co-led by Amazon, NVIDIA, and SoftBank, with post-money valuation reaching $852 billion. Top-tier investors participated, marking AI competition entry into state capital-level arms race.
OpenAI Secures $122B Funding for Global AI Infrastructure Expansion
OpenAI has raised $122 billion to expand frontier AI capabilities globally, invest in next-generation compute infrastructure, and meet growing demand for ChatGPT, Codex and enterprise AI solutions. This record funding will significantly scale up its AI training clusters and inference infrastructure.
STADLER Deploys ChatGPT at Scale to Optimize Knowledge Workflows
STADLER has deployed ChatGPT for its 650 employees, focusing on unstructured knowledge tasks processing, marking the expansion of generative AI from external customer service to internal operational efficiency optimization.
OpenAI Releases Model Spec Framework for Public AI Behavior Guidelines
OpenAI released the Model Spec framework, documenting AI model behavior rules to balance helpfulness, safety, and user autonomy. The framework is publicly available on GitHub, fostering standardized discussions on AI conduct.
OpenAI Releases Youth Safety Strategy Tool for AI Application Development
OpenAI introduces gpt-oss-safeguard, a prompt-based youth safety strategy tool providing content moderation capabilities for developers. The tool integrates safety guardrails into AI applications to address age-specific risks.
OpenAI Integrates Agentic Commerce Protocol to Expand ChatGPT's Commercial Platform Capabilities
OpenAI integrates the Agentic Commerce protocol into ChatGPT, adding visual immersive shopping features that support product discovery, side-by-side comparison, and third-party merchant integration. This embeds the complete e-commerce transaction process within the conversational AI interface, from search to purchase.
OpenAI Launches Sora 2 with Built-in Security Protection
OpenAI releases Sora 2 video generation model with native security integration at both model and platform levels. The design addresses misuse and disinformation risks through technical controls implemented at the source.
OpenAI Implements Chain-of-Thought Monitoring for AI Agent Security
OpenAI adopts chain-of-thought monitoring for internal coding agents, analyzing AI reasoning processes in real deployments to detect behavioral deviations. This approach identifies misalignment risks, marking a shift from theoretical to practical AI security monitoring.
OpenAI Launches Japan Youth Safety Governance Framework
OpenAI Japan released a customized safety framework for youth, enhancing age verification and parental controls without disclosing technical details. The scheme focuses on ethical AI governance with regional compliance strategies.
OpenAI Releases Compact Models GPT-5.4 mini/nano for Enterprise AI Inference
OpenAI launches GPT-5.4 mini and nano models optimized for coding, multimodal tasks, and high-throughput API workloads. The compact models improve inference speed and reduce deployment costs, reflecting OpenAI's strategy to enhance enterprise AI service competitiveness.
OpenAI Abandons Traditional SAST for AI Constraint Reasoning Verification
OpenAI Codex Security discards traditional SAST methods, adopting AI-driven constraint reasoning and verification to identify security vulnerabilities. This technology aims to significantly reduce false positives, representing deep innovation in AI-powered code security.
Rakuten Leverages Codex for DevOps Automation and 50% MTTR Reduction
Rakuten integrates OpenAI Codex into CI/CD pipelines for automated code review, reducing MTTR by 50% and shortening full-stack application build cycles to weeks. This demonstrates AI agents' expansion from coding assistance to operational delivery stages.
OpenAI Proposes Architectural Defense for AI Agents Against Prompt Injection
OpenAI released a technical approach implementing action constraints and data protection in AI agent workflows to defend against prompt injection attacks. The solution restricts API access and isolates sensitive data, shifting from passive filtering to proactive security architecture.
OpenAI Adds Computer Environment to Responses API for Agent Runtime
OpenAI introduces a computer environment for Responses API, integrating Shell tools and managed containers to build scalable agent runtime architecture. The architecture supports file processing, tool calling, and state management with containerized isolation for security, marking strategic evolution from base models to complete agent infrastructure.
Wayfair Applies OpenAI Models for Backend Operations Automation and Product Data Governance
Wayfair is leveraging OpenAI's model technology to automate customer support ticket routing and enhance millions of product attributes at scale. This aims to improve support efficiency and catalog accuracy, demonstrating the practical application of LLMs in enterprise backend operations automation and core data governance.
OpenAI Introduces IH-Challenge for Enhanced LLM Security Architecture
OpenAI launches IH-Challenge training technology to enhance LLM security and prompt injection resistance through instruction prioritization. This represents a shift from content filtering to underlying instruction control in model security architecture.