Tag
SiliconFlow has submitted a listing application to the Hong Kong Stock Exchange. By 2025 Token throughput, it is China's largest independent Token supplier, with investors including Alibaba, Meituan, etc.
A startup founder describes a platform enabling AI agents to pay per-call for API services, moving away from subscription models. They seek feedback on the concept.
Most MCP servers are unnecessary; this article presents a framework for deciding when an MCP server is warranted, emphasizing the need for stable APIs and CLIs first.
Google DeepMind introduces the Interactions API to pair models like Nano Banana 2 Lite for image generation and Gemini Omni Flash for animation, enabling sequential edits with session history.
Google DeepMind announced two major model releases: Nano Banana 2 Lite, a fast and cheap Gemini image model, and Gemini Omni Flash, now available via API for video generation and editing.
X has launched a hosted MCP server that allows AI tools like Claude and Cursor to easily connect to the X platform using a user's account permissions, simplifying integration and positioning X as a real-time data source for AI applications.
This article introduces AscendAPI, an API platform that provides structured, expert-verified exercise data for health and fitness apps, including images and video resources. It notes that the API's v2 version is more powerful and offers a generous free tier.
Twitter finally has an MCP interface, allowing AI to automatically organize, summarize, and analyze user data. Requires registering for Twitter API and paying per use, but personal information has a lower cost ($0.01 per call). The article details the configuration steps.
X has launched hosted MCP servers, allowing AI tools like Grok, Cursor, Claude, and VS Code to connect directly to the platform for search, bookmarks, trends, and content publishing.
Chinese developers are using gray-market API proxy services to access Claude at a fraction of the official price, exploiting account farming and free credits. These proxies pose risks of data theft and model substitution, undermining Anthropic's security measures.
Sequence is a product that enables money movement for AI agents, allowing them to handle financial transactions.
Kimi released the K2.7 Code model and its high-speed version, and announced API pricing. Compared to rival Mimo, it is more expensive and slower.
Samsung will introduce paid tiers for its SmartThings API starting October 2026, including a $4.99 monthly plan for individual developers, potentially impacting advanced users and Home Assistant integration.
A free RAG API using medical Wikipedia articles is now available to provide local LLMs with accurate medical facts, as demonstrated by correcting hallucinations about Lhermitte sign.
Cloudflare launched self-managed OAuth for all customers, allowing developers to create and manage OAuth clients for delegated API access, improving security and scalability of the Cloudflare app ecosystem.
The article discusses how LLM code style choices affect token consumption and costs, offering optimizations such as using Web API standards and simpler indentation to reduce output tokens.
A cloud lab for biology allows running experiments via API in minutes, eliminating the need for expensive automation engineers and long setup times, enabling solo founders to conduct thousands of drug screens quickly.
Databricks introduces Agent Mode API for Genie Agent, providing a new interface for building and managing AI agents on the platform.
mcpgen is a CLI tool that turns any OpenAPI 3.x spec or Postman collection into a fully functional, self-contained Python MCP server with auto-detected authentication, no runtime dependency, and generates deployable source code.
This paper introduces MemClaw, a governed shared memory architecture for multi-agent LLM systems, formalizing failure modes like unauthorized leakage and stale propagation, and evaluating the system via the ArgusFleet harness.