gpt-5

According to official news, OpenAI has officially launched the preview version of the next-generation GPT-5.6 series models, including the flagship model Sol, the balanced model Terra, and the fast low-cost model Luna. GPT-5.6 introduces a brand new maximum reasoning effort and features a super strong mode that accelerates complex tasks through sub-agents.The flagship model Sol introduces the Ultra mode, which combines maximum reasoning intensity with sub-agent collaboration. In the Terminal-Bench 2.1 command line workflow test, Sol achieved a score of 88.8%, which increased to 91.9% in Ultra mode, surpassing GPT-5.5's 83.4% and Claude Fable 5's 88.0%. The mid-range model Terra performs close to GPT-5.5 while being priced at half, and the lightest model Luna is designed specifically for everyday automation tasks. Sol is priced at $5 per million input tokens and $30 for output, and it supports reducing secondary call costs by utilizing prompt caching.In terms of security, the security assessment confirmed that Sol did not exceed the critical thresholds of the Preparedness Framework cybersecurity. OpenAI has invested over 700,000 A100 equivalent GPU hours in automated red team exercises, equipping the entire series of models with a defense stack that includes rejection mechanisms, real-time abuse classifiers, and account-level audits. Although the current limited release follows the U.S. government's security framework, OpenAI emphasizes that it does not want a government-led access mechanism to become the long-term default model, as it would limit defenders' access to cutting-edge tools.

OpenAI announced a comprehensive expansion of its cybersecurity program Daybreak, aimed at leveraging artificial intelligence to accelerate the discovery and automatic remediation of software vulnerabilities. The core of this expansion is the full version dedicated model GPT-5.5-Cyber launched for trusted defenders. It is reported that this model has set the highest score records in multiple cybersecurity benchmark tests, surpassing GPT-5.5's 81.8% and competitor Mythos 5's 83.8%, significantly improving the accuracy of vulnerability scanning and patch generation. At the same time, the synchronously updated Codex Security plugin has been deeply integrated into the developer workflow, supporting fully automated codebase scanning, threat modeling, and patch generation.In terms of ecosystem development, OpenAI has launched an exclusive partner program, allowing compliant security service providers to integrate GPT-5.5 with specific permissions into their commercial products; and has initiated the "Patch the Planet" program in collaboration with organizations like Trail of Bits to assist over 30 foundational open-source projects such as Python and Go in implementing vulnerability fixes. In addition, OpenAI revealed that it is currently engaged in deep cooperation with governments and institutions from multiple countries, including the United States, the United Kingdom, France, and Japan, to jointly enhance the cybersecurity capabilities of global critical infrastructure.

WORLD3 today announced the official launch of its decentralized AI routing network RouterLink, enabling the permanent domain routerlink.ai. This official version brings three core upgrades compared to the previous Alpha stage: complete Web2 payment support (Stripe credit card/debit card/Apple Pay/Google Pay), direct connection to over 60 cutting-edge purebred flagship models (without quantization, without distillation), and official production-level endorsements from three major cloud providers: AWS, Microsoft Azure, and Google Cloud.WORLD3 disclosed that the RouterLink network currently supports over 68 models, covering more than 20 providers, with uptime locked at 99.9%, and daily settlement flow exceeding $1.5 million $WAI. The launched purebred flagship models include Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro Preview, DeepSeek V4 Pro, Grok 4.3, Qwen3-Max, Llama 4 Maverick, Kimi K2.6, GLM 5.1, MiniMax M2.7, etc., all provided at full specifications. The cumulative $450K quota from the three major cloud collaborations has been fully utilized — including the public inclusion of the AWS Customer Case Study, Microsoft for Startups Level 4 (highest level) + $300K AI Grants, and direct connection to Google Cloud Vertex AI.WORLD3 stated that this official launch also opens up the enterprise procurement channel: Stripe payments and $WAI on-chain payments run in parallel, allowing developers to access the complete catalog of 68 models within 60 seconds after registering their email and recharging with a credit card. The subsequent roadmap includes the RouterLink mainnet, GPU computing power market, an Agent skill market linked with TapClaw, and an enterprise version SLA service that is already in pilot with some clients.

The B.AI API model library welcomes a significant expansion, with four major models officially launched: GPT-5.5 Instant, DeepSeek-v3.2, MiniMax-M2.7, and GLM-5.1.Among them, GPT-5.5 Instant completed underlying adaptation and interface integration within 48 hours after OpenAI's release, achieving zero-latency calls for the new generation of computing power. Compared to the previous generation GPT-5.3 Instant, the latest model GPT-5.5 Instant has reduced hallucination rates by 52.5% in high-precision scenarios such as healthcare, law, and finance, significantly improving logical clarity and output conciseness, greatly enhancing the factual reliability of complex tasks.Currently, the above models are fully open for B.AI API requests. Developers can refer to the official documentation to quickly access the latest model interfaces, providing a more stable and trustworthy AI capability with zero latency experience.

According to official news, B.AI officially completed the full integration and deployment of the next-generation large models GPT-5.5 and Claude Opus 4.7 on May 1. Developers and users can directly experience these two industry-leading models through https://chat.b.ai/chat. GPT-5.5 has achieved significant advancements in reasoning ability, contextual understanding, and multimodal interaction, while Claude Opus 4.7 excels particularly in code generation, logical alignment, and complex task execution. This model upgrade aims to continuously provide cutting-edge AI computing power support for users within the ecosystem, lower the threshold for developers to access top models, and ensure higher response speed and stability.

According to official news, OpenAI has officially released GPT-5.5. This model has made significant progress in intelligent agent programming, computer usage, and scientific research, achieving an accuracy of 82.7% in the Terminal-Bench 2.0 test. GPT-5.5 has higher reasoning efficiency and consumes fewer tokens while maintaining the same latency as GPT-5.4.Currently, GPT-5.5 has been pushed to ChatGPT Plus, Pro, and enterprise users, with the API version set to launch soon, priced at $5 per million input tokens and $30 per million output tokens. Additionally, GPT-5.5 Pro, designed for high-difficulty tasks, has also been released simultaneously.

OpenAI announced the launch of the GPT-5.4-Cyber model, which is fine-tuned from GPT-5.4 for cybersecurity scenarios, reducing the refusal limits on security-related requests under specific conditions to support specialized operations such as binary reverse engineering.According to the introduction, the model is only available to vetted security vendors, corporate security teams, and researchers through the "Trusted Access for Cybersecurity" (TAC) program. OpenAI stated that the relevant layered mechanism will limit the scope of model usage and impose additional constraints on access in low visibility scenarios.

OpenAI announced the release of GPT-5.4 in ChatGPT (in the form of GPT-5.4 Thinking), API, and Codex, marking its most powerful and efficient state-of-the-art model for professional work to date. It also released GPT-5.4 Pro for ChatGPT and API users, designed specifically for those who seek extreme performance on complex tasks.GPT-5.4 integrates OpenAI's recent advancements in reasoning, programming, and agent-based workflows into a single state-of-the-art model, incorporating the industry-leading programming capabilities of GPT-5.3-Codex while improving the model's performance in tools, software environments, and professional tasks involving spreadsheets, presentations, and documents.

OpenAI today released the GPT-5.3-Codex-Spark research preview, a streamlined version of GPT-5.3-Codex, and the first model designed specifically for real-time coding, supporting targeted editing, reshaping logic, or optimizing interfaces, with immediate result visibility.Codex-Spark has been optimized to run on ultra-low latency hardware, achieving near-instantaneous responses—delivering over 1000 tokens per second while still handling actual coding tasks.

OpenAI today announced the launch of GPT-5.2-Codex, based on GPT-5.2. OpenAI co-founder Sam Altman stated that it has been specifically trained for agent coding and end-user applications, and OpenAI employees have achieved great success using it.Additionally, OpenAI is reportedly planning to raise up to $10 billion at a valuation of $830 billion.

OpenAI has launched GPT-5.2, focusing on professional knowledge work and long-process AI agent design. GPT-5.2 offers three versions: Instant, Thinking, and Pro. Instant is suitable for writing and information search; Thinking excels in deep work such as program development and planning; Pro provides the most accurate responses for complex problems.OpenAI states that the core goal of GPT-5.2 is to make AI more "economically valuable" in real work scenarios. Additionally, in multiple benchmark tests, GPT-5.2 outperformed industry professionals in various aspects.

According to OpenAI's announcement, GPT-5.1 has begun a full rollout to all Chat GPT users this week. This version shows significant improvements in following custom instructions, reasoning accuracy, and conversational naturalness. GPT-5.1 supports adaptive reasoning, intelligently extending thinking time for complex questions, resulting in clearer responses with fewer technical terms. GPT-5 users can continue to use it for three months, and the Pro version will be updated to GPT-5.1 Pro.

ChainCatcher news, according to official sources, OpenAI officially launched GPT-5 on August 7, showcasing groundbreaking performance in programming and complex task handling, further advancing the development of AGI (Artificial General Intelligence).However, with the power consumption of Nvidia Rubin chips soaring to 600kW per rack, the contradiction between the rapid increase in AI computing power and the global energy crisis is becoming increasingly acute. Technical experts warn that if the energy efficiency bottleneck cannot be overcome, the AI industry may face a "power cliff."SunPump will host a roundtable discussion titled "GPT-5 × Energy Crisis" on Twitter Space at 8 PM on August 19. This discussion will bring together industry leaders to deeply analyze the race for survival between intelligent evolution and sustainable energy in the era of GPT-5, covering technical pathways, economic costs, and geopolitical competition. This dialogue may redefine the future boundaries of AI.The seminar will be broadcast live via Twitter Space. Users can follow the official accounts @sunpumpmeme and @Agent_SunGenX, retweet the event tweet, and @ three friends to participate in an interactive lottery for a chance to win a reward of 10 USDT.

ChainCatcher news, according to Jin10 data reports, despite the launch of the new GPT-5 model, ChatGPT still has the potential to make errors. A senior executive from OpenAI reiterated this point again this week. Nick Thorne, the head of ChatGPT, pointed out in an interview: there is a significant discontinuity between being reliable and being completely reliable.He further explained: unless we can prove that ChatGPT is more reliable than human experts in all areas, not just certain areas, we will continue to advise you to double-check the answers.

ChainCatcher news, according to Zhitong Finance, CITIC Construction Investment has released a research report stating that OpenAI has officially launched GPT-5, which includes three versions: GPT-5, GPT-5-Mini, and GPT-5-Nano.GPT-5 excels in long text memory, reduced hallucination rates, and optimized reasoning efficiency, providing a better foundation for the development of AI applications. Future updates from Google and Renze, as well as the launch of domestic models, can be expected, along with the continued implementation of text AI application tokenization under model optimization.In addition, Huawei has fully open-sourced the CANN ecosystem, diversifying its offerings for different network developers, significantly improving development efficiency, and is expected to accelerate the catch-up with CUDA, providing a better foundation for domestic AI application development.

ChainCatcher news, Tianyancha App shows that OpenAI, Inc. has applied to register two "OPENAI GPT-5" trademarks, classified internationally as website services and scientific instruments. The status of these trademarks is currently pending appeal.In addition, OpenAI OpCo, LLC has applied to register two "GPT-5" trademarks, classified internationally as website services and scientific instruments. The status of these trademarks is currently under appeal. (Interface)

ChainCatcher news, according to Jinshi reports, OpenAI has launched a more powerful GPT-5 model for programming and writing.

ChainCatcher news, OpenAI officially announced on the X platform that it will hold a live event on Thursday at 10 AM Pacific Time (1 AM Beijing Time on Friday).The official tweet replaced the "S" in "LIVESTREAM" with a "5", sparking speculation about the GPT-5 model.

ChainCatcher news, according to a post on the social media platform "X", OpenAI CEO Sam Altman stated: "In the near future, an artificial intelligence smarter than the smartest person you know will run on a device in your pocket, helping you accomplish anything you want to do. This is an incredible thing."Recently, multiple media outlets reported that OpenAI's next-generation flagship model GPT-5 is set to be released, and OpenAI executives have also been frequently signaling related information on social media. (Jinshi)