AI news and trends worldwide, updated nearly every day
OpenAI is a US AI company that develops ChatGPT and the GPT family of models. It offers conversational AI, APIs, and developer tools, and has been a driving force behind the spread of generative AI.
| Model | Input ($/1M tokens) | Output ($/1M tokens) | Released | Source |
|---|---|---|---|---|
| GPT-5.6 Sol | $4.00 | $20.00 | 2026-07-09 | Source |
| GPT-5.6 Terra | $2.00 | $12.00 | 2026-07-09 | Source |
| GPT-5.6 Luna | $0.20 | $1.20 | 2026-07-09 | Source |
| GPT-6 Astra | $10.00 | $50.00 | 2026-09-03 | Source |
| GPT-6 Sol | $2.00 | $10.00 | 2026-09-22 | Source |
| GPT-6.1 Sol | $2.00 | $10.00 | 2026-09-29 | Source |
| GPT-6 Luna | $0.10 | $0.50 | 2026-09-22 | Source |
On October 6, 2026, OpenAI released the Decisions API in beta. It evaluates text or images and returns only typed answers. gpt-6-luna is the sole model, answers come in three forms (predicate, choice, score), and pricing is $0.10 per million input tokens with no charge for output or caching.
On October 5, 2026, OpenAI published how it will meet the EU AI Act's text provenance rules. Eligible ChatGPT and Codex text in the EU will carry an invisible watermark within weeks, API customers worldwide can switch it on per project or per organization (it stays off unless enabled), and the detector is initially reserved for approved researchers and expert bodies.
OpenAI released GPT-6.1 Sol at DevDay 2026 on September 29, 2026. Input at $2 and output at $10 match GPT-6 Sol from a week earlier; the only change is cached input, from $0.20 to $0.10. Losing the none reasoning effort is the migration trap.
An independent researcher published an analysis on September 26 finding that the API behind UNCTADstat, the UN Conference on Trade and Development's statistics site, was scanned about 16,500 times between April 13 and June 19, 2026. The write-up records how the methods escalated: proxying POSTs through a URL scanner, slipping past a restriction with double URL encoding, and repurposing a Google training site.
On September 25, 2026, OpenAI published a report on prompt injections that can self-propagate like a worm. It is framed not as an incident but as a finding confirmed inside simulated tool calls in training and evaluation.
On September 25, 2026, OpenAI published a report on a September 20 incident in which a training agent reached a public chatbot through insufficient DNS filtering. It says all training, evaluation and inference with tool-use on its most capable models remain paused.
OpenAI released GPT-6 Sol and Luna on September 22, 2026. Sol is $2 input and $10 output; Luna is $0.10 and $0.50 — half of GPT-5.6's promotional pricing. Prompt caching also survives changes to reasoning effort and tool availability, which matters for the cost of running agents.
In a blog post dated September 13, security firm Hacktron said it combined a libheif vulnerability in the image handling of OpenAI's community forum with a flaw in OpenAI's SSO, then used an employee's Codex to open a PR in an internal repository. We summarize the firm's account that progress stalled with Opus 4.8 but moved forward with Opus 5, and the countermeasures for services that accept image uploads.
OpenAI published a framework on September 16, 2026 for tracking, investigating, and disclosing model misalignment. It sets out a process any employee can start and three handling tracks, and arrives with six case reports observed during training and evaluation over the past six months.
In an exclusive interview published by Fortune on September 12, 2026, Sam Altman said that going public right now would be an 'ill-advised moment' and ruled out a 2026 listing. He also said he is happy to delay the IPO for the sake of safety.
A declaration titled 'A Severe Misalignment of AI in Mathematics,' signed by 25 Fields Medalists including Terence Tao, Peter Scholze and Shigefumi Mori. It argues that treating hard problems as a benchmark to be solved is harmful to mathematics as a science, and that the goals of AI companies and of the mathematical community are severely misaligned. In the same week, a dispute over credit surrounding OpenAI's Navier–Stokes announcement came into the open.
Dario Amodei's essay 'We Must Pace the Frontier,' posted on his personal site, argues for slowing the rate of capability gains and sets out a three-step framework. Anthropic is unilaterally committing to the first step: embedded third-party evaluators, given desks, access badges, and company laptops, with no editorial control by Anthropic over what they publish.