Reader-focused coverage of AI models, APIs, infrastructure, developer tools and changes that matter in practice. Client service announcements remain inside the authenticated account area.
OpenRouter’s new Batch API lets teams submit latency-tolerant LLM work across chat, responses, messages, and embeddings endpoints, with provider discounts generally tied to 24-hour completion windows.
Read news →OpenAI now lets eligible Enterprise and Edu admins assign identity-provider groups to API organizations and projects, reducing manual invites and tightening access governance for API teams.
Read news →Microsoft’s latest Azure API Management AI Gateway update adds model aliases, Entra-based access, gateway-scoped policy controls, and forced usage capture for streamed OpenAI-compatible responses.
Read news →Claude Opus 5.5 is now available through Anthropic’s API and major cloud platforms, with lower published token prices and a larger context profile for long-running coding and agent workloads.
Read news →OpenAI has released GPT-6 Sol and GPT-6 Luna to its API, with same-day availability across major cloud and developer gateways and pricing that sharpens the case for workload-based model routing.
Read news →Grok 4.7 is now available through the xAI API, Vercel AI Gateway, GitHub Copilot, and other developer surfaces, giving teams another frontier coding and agent model to test under routing, cost, and governance controls.
Read news →Z.ai’s GLM 5.3 FlashX is now available through Vercel AI Gateway, adding another low-cost, OpenAI-compatible option for coding agents, tool loops, and latency-sensitive multimodal applications.
Read news →GitHub Copilot now lets teams steer automatic model selection toward efficiency, balance, or intelligence, turning coding-assistant model routing into a more explicit cost and quality governance decision.
Read news →Alibaba’s Qwen3.8-Omni-Flash gives developers an OpenAI-compatible multimodal API model with a 1 million-token context window and unusually low listed prices, sharpening the routing and cost-control tradeoffs for teams handling text, image, audio, and video inputs.
Read news →Cloudflare AI Gateway can now require customer-supplied provider credentials, giving teams a way to prevent fallback to Cloudflare-managed Unified Billing when stored keys are missing.
Read news →Google’s Gemini 3.8 Live release gives developers a generally available audio-to-audio model for real-time voice agents, with per-minute pricing that makes voice workloads easier to compare, route, and govern.
Read news →Google Cloud API Gateway can now act as a remote MCP server in public preview, giving teams a way to expose existing REST APIs to AI agents without rebuilding backend services.
Read news →