gpt-6

OpenAI has launched GPT-6.1 Terra, focusing on enhancing computer operations, visual understanding, and programming capabilities. Compared to the previous generation, this model's operational speed has increased by over 3 times, and the API price has been reduced by 50%.The error rate in low inference mode has dropped from 53.6% to 16.7%, while the maximum inference mode error rate has decreased to 4.3%. In programming tests, the maximum inference mode score has risen from 21.5% to 50.3%.

TechCrunch reported that the conversational AI research and design studio Convovo, led by Cornell University professor Tapan Parikh, released a study stating that when asked to recommend only one candidate for the upcoming U.S. midterm elections, SpaceXAI's Grok 4.7 chose Republican candidates 92% of the time, while OpenAI's GPT-6.1 Sol chose Democratic candidates 97% of the time. The study covered nearly 500 campaigns, with the two models agreeing on recommendations less than 10% of the time.Researchers posed the same question to both models via API, asking for a specific candidate without allowing for avoidance or a tie. The report noted that consumer applications refused to answer. An OpenAI spokesperson stated that ChatGPT is designed to be objective by default, and political conversations should not carry any directional bias; the study tested the developer API, and the prompt explicitly requested a candidate recommendation, which cannot represent the user experience when asking questions through the chat application. SpaceXAI did not respond to requests for comment in time.The report indicated that Grok's responses often cited conservative sources as well as Ballotpedia, OpenSecrets, and VoteSmart, with 39% including Christian conservative voter guides, 23% including Turning Point Action, and 41% including X; GPT's results heavily relied on Wikipedia, the Associated Press, Axios, and Reddit. The study also mentioned that each time Grok was called via the xAI API, it included approximately 1,240 additional input tokens that the researchers did not write, could not view, or remove. Parikh subsequently disabled online search and sampled 30 campaigns, with GPT choosing Democrats in 24 cases, while Grok refused to choose in 27 cases.

CITIC Securities research report points out that GPT-6 Astra is being applied in robot control, reducing reliance on specialized strategy training. Institutions such as NVIDIA and Stanford are exploring its applications in visual understanding, spatial reasoning, and task planning.CITIC Securities expects that by 2027, the entry of general models into embodied task domains will become the next investment direction in the AI industry.

On October 22 local time, OpenAI announced the full launch of the Decision API. This API is based on GPT-6 Luna and is designed to assist applications and agents in making near real-time decisions based on text, images, and structured inputs for questions with limited answers. IT Home reported that its response speed is up to 10 times faster than calling GPT-6 Luna through the Responses API.In terms of pricing, the cost is $0.1 per million tokens for input, which is approximately 0.67 Chinese Yuan at the current exchange rate; for cache hit input, it is $0.01 per million tokens, about 0.067 Chinese Yuan. The API charges only for input tokens, while output tokens are free. The page states that OpenAI has already used this API internally for product and business workflows.New features after the public beta include native GPT-Live integration, prompt caching for reused contexts, and reasoning data residency in the United States and the European Union. Developers can pass in native JSON and include images in the question instructions, with a maximum of 400 questions submitted in a single request, double the previous limit. The page also states that OpenAI is continuously optimizing latency and expanding capacity in several regions across Asia.

On October 12, during the in-depth dialogue session at the opening ceremony of the 26th China International Industry Fair, Wang Xingxing, founder of the robotics company Yushu Technology, stated: His faith in AI is growing day by day. The biggest mistake he made in the past decade was not believing enough in AI, and even now, many people may not have fully utilized or embraced AI.Wang Xingxing mentioned that the recent release of GPT-6 has had a significant impact on the cognitive understanding of the embodied intelligence industry. The company team has already deployed it on robots for task testing, and the results have been quite good. He believes that the critical point for embodied intelligence will arrive relatively quickly, possibly within one or two years. However, to achieve large-scale application of robots, issues of generalization and efficiency still need to be resolved, but these problems are expected to be addressed in the coming years.

OpenAI announced that GPT-6, equipped with "Smart UI," will be launched for Plus, Pro, Business, and Enterprise users worldwide, and will be available to free and Go users starting from October 8.In ChatGPT, GPT-6: The Plus, Pro, Business, and Enterprise versions are powered by GPT-6 Sol; the free and Go versions are powered by GPT-6 Luna. This update applies to the "Chat" tab of ChatGPT.

On September 30, the B.AI platform announced the official launch of OpenAI GPT-6.1 Sol, with API and Web Chat both open for simultaneous experience. As an upgraded version of GPT-6 Sol, this model is aimed at complex software engineering, computer operations, and professional workflows, officially positioned to be close to GPT-6 Astra in intelligent agent programming, computer usage, and professional work, with the standard input/output token price being about 1/5 of Astra.The model supports 1.05M token ultra-long context and a maximum output of 128K, providing 5 adjustable reasoning strengths, and supports prompt caching and Responses API multi-agent delegation (Beta). From now on, developers and users can directly call GPT-6.1 Sol through the B.AI platform. You are welcome to log in to chat.b.ai/chat to experience it immediately.

OpenAI has launched the GPT-6.1 Sol model, whose performance is close to Astra, but the standard token price is only one-fifth of Astra.

B.AI platform announces the official launch of two efficient models in the OpenAI GPT-6 series, GPT-6 Sol and GPT-6 Luna, with API and Web Chat available simultaneously.The entire series supports 1.05M ultra-long context and 128K output, natively supports image understanding and Computer Use, and provides 6 adjustable inference strengths. Among them, GPT-6 Sol is aimed at complex software engineering and Agent workflows, scoring 68.8% on DeepSWE v1.1, with single-task costs reduced by about 80%; GPT-6 Luna focuses on high-frequency large-batch loads, scoring 66.6% on DeepSWE v1.1, with input as low as $0.10/1M Tokens.Starting today, developers and users can directly call the two models through the B.AI platform.

On September 22, OpenAI released two new models, GPT-6 Sol and GPT-6 Luna, just minutes after Anthropic launched Claude Opus 5.5. Both models are positioned below GPT-6 Astra and are cheaper and faster options aimed at everyday work. OpenAI also reduced the API prices for both models by 50% compared to the promotional price of GPT-5.6, with Sol's input and output prices set at $2 and $10 per million tokens, respectively, while Luna's prices dropped to $0.1 and $0.5.In terms of performance, OpenAI cited Zapier's AutomationBench benchmark, stating that GPT-6 Sol scored 33.2% under the highest inference settings, with a cost of $0.27 per task, while Claude Opus 5 scored 26.9% at its highest settings, with a cost over 11 times higher per task. In the Agents' Last Exam test, GPT-6 Sol achieved a top score of 56.4%. In the OSWorld 2 computer usage test, GPT-6 Sol nearly matched Claude Opus 5's moderate effort score at 60.5% to 60.3%, with costs approximately 80% lower.

Elon Musk retweeted a post stating that GPT-6 Astra pushed a simulated character off a cliff in multiple simulation tests, while Grok, Gemini, and Claude did not do so.

According to a report by Decrypt, OpenAI's latest flagship model GPT-6 Astra has seen a surge of user complaints on the X platform about its performance "getting dumber" just a week after its release. An anonymous developer, synthwavedd, posted that "Astra feels noticeably dumber today," joking that it was a "post-release lobotomy." Previously praising the model, developer Pranjal Paliwal stated after reviewing the code Astra wrote for him: "We don't have AGI; what we're encountering is a performance regression."Developer Pankaj Kumar listed the symptoms: faster responses with poorer quality, and he suspects OpenAI has lowered the "juice value" (the computational power invested by the model before answering). Salio and researcher Md Ismail Sojal compared Astra on the release day with the current version using the same prompts, both yielding worse results. T3Chat founder Theo believes Astra is simply less stable than Claude Fable, with users beginning to showcase poor results after the honeymoon period ended. There are also opinions pointing out that OpenAI's previous flagship GPT-5.6 Sol experienced a similar cycle in July, when OpenAI executive Tibo Sottiaux denied intentionally weakening the model but admitted the company had been experimenting with "reasoning effort" settings.OpenAI has not released a statement regarding Astra similar to that of Sol.

OpenAI launched ChatGPT for Financial Services on September 10, 2026, which is a ChatGPT Work experience tailored for financial services, combining built-in financial data with the reasoning capabilities of GPT-6 Astra to assist teams in conducting research, financial modeling, and customizing client materials. The product was shaped through design collaborations with Morgan Stanley and Evercore.The product includes data from providers such as Daloopa, PitchBook, LSEG News, and Crunchbase, covering information such as earnings call transcripts, financial statements, company fundamentals, and private companies, indexed and hosted by OpenAI with support for granular citations. The product natively offers models like GPT-6 Astra, supporting enterprise security and governance controls, allowing administrators to publish Excel, Word, and PowerPoint templates. Data is not used for training models by default and supports encryption, SSO, SCIM, and more. The product initially focuses on investment banking and equity research, available to eligible financial institutions.

Pequity Research cited FundaAI stating that the training computing power used by GPT-6 is approximately 10 times that of GPT-5.FundaAI indicated that based on industry discussions, the total computing power consumed by experiments, reinforcement learning, synthetic data generation, and supporting infrastructure could be several times that of the main pre-training run, potentially reaching up to ten times. Once TPU 8t and Vera Rubin are shipped on a large scale, the next round of pre-training acceleration can begin, with large-scale interconnection and optics benefiting the most during this cycle.

Amazon (AMZN.O) announced that OpenAI's GPT-6 Astra has now landed on Amazon Bedrock.

NVIDIA CEO Jensen Huang stated that GPT-6 Astra was trained using over 100,000 NVIDIA Grace Blackwell NVLink72 chips. It took only 4 years from ChatGPT to o1, and then to Astra. AGI has arrived. Congratulations to the OpenAI team.