API

On October 27, OpenAI published GPT-6.1 Tera, which focuses on visual understanding, computer operations and natural writing, and has been made available to users of the ChutGPTwork and Codex Plus, Pro, Enterprise and Edu, who can call through API. OpenAI indicates that the new model standard API prices are lower than GPT-5.6 Terra and GPT-6.1 Sol 50%: Token is 1 dollar per million..

On October 20, OpenAI ' s Codex and ChatGPT team member Tibo Sottiaux stated that the team had worked with OpenCode to help restore service to affected users who were also using OpenAI API and to return the API amount consumed during the attack. OpenCode has publicly disclosed that a security incident related to a large-scale model distillation attack affected the Console user of 1.4%, with the range of exposure at cloud level..

On September 23rd, OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, decentralizing part of GPT-6 Astra ' s progress in professional work, factual accuracy, programming and computer operations to two models that are faster and cheaper. Officially, the API prices of both new models were reduced by 50% compared to the GPT-5.6 promotion price. GPT-6 Sol Input and Output Prices per 1 million Token

On July 9th, SpaceXAI officially released Grok 4.5, which is positioned as the firm ' s strongest model to date, with the lead code, AI smart body mission and knowledge-based work. The model was combined with Cursor training based on a new V9 base, with parameters of approximately 1.5 trillion, using tens of thousands of British GB300 GPUs. In performance, Grok 4.5 outperformed Claude Opus 4.8..

On June 15th, OpenRouter, a model synthesizing platform, released Fusion, where users can send the same hint to a set of models in parallel, like a single model, and then analyse consensus, contradictions, blind areas and unique information by a judice model, and eventually synthesize the answers. Officially, 100 missions were tested on the DRACO in-depth study benchmark. Budget panel comprising Gemini 3 Flash, Kimi K2.6 and DeepSeek V4 Pro scored 64.7%, close to Fab..

reAPI is an AI API aggregation platform that provides access to all top AI models through a single endpoint. It supports the generation of images, videos, chats, music and codes and ensures the normal running time of 99.96%. ReAPI functionality Aligns API endpoints: All top-level models can be called through an API key and OpenAI compatible interface, without having to be integrated for each supplier. Automatic failure transfer: Each popular model is supported by multiple suppliers, and when a supplier is downgraded, an automatic transfer to a back-up route is requested to apply undetected. Zero Request Log: Request and..

In recent days, the Director of the AI "API Transit Station" station in Shanghai has been on the Internet and has been detained for 37 days in criminal custody on suspicion of selling offshore AI interface resources, which has since been changed to bail and is still under investigation. As soon as news emerged, a large number of similar platforms were shut down and moved, and the whole industry was booming. At present, there are rumours that the case may be prosecuted for illegal dealing, so what is the space for defence? What's the business at the stop? A brief description of the operational modalities of such platforms. Because the main models of OpenAI, Claude and others have made land for mainland China..

On May 26th, the dark side of the moon, the official Kimi announced last night that Kimi K2 model API was offline. The kimi-k2 series model will be offline on 25 May 2026 and will no longer be maintained and supported. Please use the Kimi 's latest model kimi-k2.6 directly to obtain continued support and more robust multimodular understanding, reflection, code and Agent capabilities. Kimi-k2 series models include: kimi-k2-0711-preview kimi-k2-..

On May 23rd, DeepSeek officially announced that the current DeepSeek-V4-Pro API discount of 2.5 would be converted to a permanent price reduction at 23:59 on May 31st. As of 1 June, the model API was adjusted to a quarter of the original price. DeepSeek API documents show that V4-Pro currently supports a 1M context, enter a cache outstanding price of $ 0.435 per million tokens and output price of $0.00..

SERIOUSLY, I'VE HAD ENOUGH! THERE ARE 78 OF THEM, AND NONE OF THEM SAVES ME. CHEAP FEAR OF PITS — USING GLM TO IMPERSONATE GPT-4 GILLS. THE EXPENSIVE ONES DON'T LOOK GOOD EITHER. THEY'RE OFTEN BROKEN. THE MOST RIDICULOUS THING IS THAT IT TOOK A FAMILY THREE MONTHS TO MAKE A SUDDEN RUN, THAT MONEY WENT STRAIGHT TO THE WATER AND THAT IT WAS IMPOSSIBLE TO FIND A PASSENGER. I TOOK THE TABLE UP: ISN'T THAT THE STOP? I'LL DO IT MYSELF! TODAY I'M GOING TO CLEAN UP THE WHOLE PROCESS OF BUILDING A STOP FROM ZERO AND STOP LOOKING AT THE POOR MERCHANTS! 📚 CATALOGUE, POSSIBLE RISK TO THIRD PARTY TRANSIT, ENVIRONMENTAL READINESS! EVERYONE..

ON NOVEMBER 13TH, AT TODAY'S 2025-DEGREE WORLD CONGRESS, THE 100-DEGREE FOUNDER, LI JIANG HONG, STATED THAT THE VAST MAJORITY OF THE SEARCH RESULTS WERE GENERATED BY AI, WHILE THE FIRST RESULT HAD A MEDIA COVERAGE OF 701 TP3T. "YOU NOW SEARCH FOR 10 QUESTIONS, 7 ANSWERS ARE RICH MEDIA, NOT A TEXT LINK, BUT A PICTURE, A VIDEO, A LIVE BROADCAST, EVEN A DIGITAL PERSON." ACCORDING TO LI, 100 DEGREES IS THE MOST RADICAL OF ALL SEARCH ENGINES AROUND THE WORLD. RECONSTRUCT THE SEARCH RESULTS PAGE WITH 100 DEGREES OF AI, NOT SIMPLY SEARCHING FOR KNOTS..

On September 29th, DeepSeek officially released today the DeepSeek-V3.2-Exp model, an experimental version. As an intermediate step towards a new generation of structures, the V3.2-Exp introduced DeepSeek Sparse Attention (Note: A Rare Focusing Mechanism) on the basis of V3.1-Terminus, to explore optimization and validation of the training and reasoning efficiency of long texts. DeepSeek Spa..

What are the excellent AI big language models in China that may equal ChatGPT? In the process of learning AI large language models, Pothead has compiled study notes on 12 domestic large language models, including the name of the large language model, the research and development team, access to the link, a partial list of models, and a brief introduction. Because the model content and price of the major company teams will be different at different times, it is specifically stated that this paper is recorded on July 19, 2025. ▌Moonshot Official Website https://www.moonshot.cn/ Introduction The language model used by Kimi's intelligent assistant...

May 22, 2012 - OpenAI Inc. published a blog post yesterday (May 21) announcing the expansion of the Responses API to support remote connection to MCP servers, image generation, Code Interpreter, and file search optimization to help developers build smarter intelligent body applications. 1AI queried the public information, Responses API is OpenAI's stateful API, supporting a variety of new tools including web search, file search, and computer usage, providing developers with more concise...

May 17, 2011 - Technology media outlet Android Authority published a blog post yesterday (May 16) reporting that Google will introduce new ML Kit GenAI APIs to extend access to the end-side AI model Gemini Nano. Google is expected to introduce these APIs in detail at its I/O developer conference in 2025, where developers will be able to integrate features such as text summarization, proofreading, rewriting, and generating image descriptions in their apps using these APIs. 1AI cites a blog post that describes how, compared to...

May 12, 2011 - Star AI company Anthropic announced the launch of web search functionality for its API on May 8, local time, meaning that Claude AI has been able to access the latest information on the web, putting new pressure on the Google search engine, which was born in the era of traditional search. Developers utilizing the Anthropic API can now enable web search tools when sending requests, broadening Claude's knowledge with current data from the real world. This also means that developers can use it to build insights that provide the latest...

May 6, 2011 - AI company Anthropic today announced the launch of its AI for Science science support program. Through this program, selected researchers will receive free and substantial API credits to accelerate scientific research and discovery. Anthropic believes that AI has the potential to dramatically accelerate scientific progress: advanced AI reasoning and language capabilities can help researchers analyze complex data, generate hypotheses, design experiments, and communicate discoveries more efficiently. anthropic's AI technology reduces the...

May 2 - Technology media outlet KrebsOnSecurity published a blog post yesterday (May 1) reporting that an employee of Elon Musk's artificial intelligence company xAI had inadvertently leaked an API key on GitHub, a mistake that lasted nearly two months. Philippe Caturegli, the "Chief Hacking Officer" of French security consultancy Seralys, was the first to expose the issue on LinkedIn, and security firm Git...

April 24, 2011 - OpenAI announced on Wednesday that it is bringing the technology behind its new and upgraded image generation feature in ChatGPT to its API, enabling developers to integrate the feature into their apps and services. OpenAI's new ChatGPT image generator, launched in late March, quickly gained popularity for its ability to create photorealistic Ghibli-style images and "AI dolls". The launch attracted millions of new users to sign up for ChatGPT and put a strain on the company's server capacity. According to Open...

April 15, 2011 - OpenAI today announced that it will soon stop making its largest AI model to date, GPT-4.5, which was just released in late February, available through its application programming interface (API). OpenAI said developers can continue to use GPT-4.5 through the API until July 14th. After that, developers will need to move on to other models in the OpenAI catalog. The company also unveiled the new GPT-4.1 model today, positioning it as the GP...

April 10, 2011 - Musk's artificial intelligence startup xAI announced its latest flagship model, Grok 3, on February 17, but only now has it launched its corresponding API interfaces, including a streamlined version, the Grok 3 Mini. Pricing for the series of products is benchmarked against Anthropic's Claude 3.7 Sonnet, with Grok 3 priced at $3 and $15 per million words (token) input/output, respectively, and an accelerated version priced at $5.... (token) input/output pricing is $3 and $15 per million words, respectively, with the accelerated version priced at $5...

Elon Musk's artificial intelligence company xAI has added image generation to its API. The API only supports the "grok-2-image-1212" model, which generates up to 10 images based on input descriptions, handles up to 5 requests per second, and produces output in JPG format, charging $0.07 per image (note: the current exchange rate is about $0.51 yuan). xAI partnered with AI startup Black Forest Labs last year to push the model on Musk's social media platform X...

March 20, OpenAI yesterday (March 19) in the X platform tweet, announced through the developer API, officially launched the o1 series upgrade "o1-pro", claiming that it through a higher investment in computing resources to achieve "more consistent and high-quality response! The core advantage of o1-pro is that it provides "more consistent and better quality responses" through higher investment of computing resources. The core advantage of the o1-pro is a significant increase in computational resources, allowing it to more easily navigate complex problems, but the new model is only available to select developers (who need to spend at least $5 on the API service) and is priced well above similar products. According to the official pricing...

March 9 news, the national supercomputing Internet platform announced yesterday evening that this week on-line Alibaba's open source inference model QwQ-32B API interface service, users can get free 1 million Tokens. QwQ-32B is the latest release of Ali Qwen team's inference model, based on the construction of Qwen2.5-32B + Reinforcement Learning. According to the official benchmark results, QwQ-32B performs well in the AIME24 test set, which tests math ability, and LiveCodeBench, which evaluates code ability....