OpenAI's GPT 5.6 Model Surfaces in Cursor's Internal Preview
Following OpenAI's announcement of GPT-5.6 Sol, its flagship language model, a sighting on X indicates the software has already appeared in the internal testing environment of AI coding platform Cursor. Multiple users posted links showing the model available for preview testing within the application, roughly a week before the company's planned July rollout.
The sighting aligns with OpenAI's phased release for the GPT-5.6 series, which includes the faster Luna tier and the balanced Terra tier alongside the highly capable Sol version. Sol is currently restricted to a narrow group of government-approved partners and enterprise customers under a new US regulatory framework, with Cursor likely among the early beta testers. The model will expand to Cerebras semiconductor infrastructure as scheduled next month.
From the sources (25 posts)
@kaicathycRT @MicahCarroll: GPT-5.6 Sol is a significant step up in capabilities, but can also exhibit concerning forms of misaligned behaviors in ag…
@theoGPT-5.6 is here. I wish we could use it.
@stablequanRT @AnthropicAI: Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today,…
@calilyliuWelcome to our most recent nCMO. We agree that token capital efficiency is paramount.
@mtsliveDAILY SITUATION RECAP: GPT-5.6 is released in limited preview. 5.6 is OpenAI’s answer to Mythos/Fable: similarly capable, with similarly robust safety filters. It has much-improved agentic capabilities in bio and cyber, exceeds Mythos on
@mtsliveThe US gov allows Anthropic to release Mythos to 100+ major companies and government agencies, including their foreign national employees, a good sign for the state of US-Anthropic relations and for the broader return of Fable 5.
@rohanpaul_aiOpenAI wrote in their GPT-5.6 official blog post today. On Trump administration's selective approval process of new model release.
@reutersUS allows Anthropic to release Mythos AI to 'trusted' US organizations
@francoisfleuretReading the news is like reading a Gibson novel from the 80s.
@ns123abc🚨U.S. Commerce Sec. Howard Lutnick’s official letter to Anthropic's COO Tom Brown MYTHOS 5 ban lifted... ONLY for: - Anthropic’s own non-U.S citizen researchers - US "trusted partners" + their foreign-national employees - US government: ci
@blackroomsecIf you have been worshiping at the altar of these morons, please stop debasing yourself. Really, stop now Us tech experts have tried to tell you, you were going to be played and now you have been. Learn from this experience. No, learn f
@thehackersnews⚡ OpenAI is keeping GPT-5.6 behind a narrow gate. Sol, Terra, and Luna are in limited preview for government-approved partners. Sol adds stronger cyber safeguards for #vulnerability research, defensive testing, and dual-use requests. Rea
@thehackersnewsEngineers spent several weeks pressure-testing the flagship Sol architecture to build its most robust deployment safety stack to date. The system actively monitors incoming tokens to block offensive cyber tactics and refuse what OpenAI cla
@thehackersnewsSecurity research teams validated the model's advanced vulnerability discovery and patch generation capabilities using the ExploitBench evaluation harness.
@thehackersnewsThe rolling deployment requires stringent case-by-case authorization under active policy pressure and security review from the U.S. government. Learn more -
@financialjuiceTrump administration near to let Anthropic regain access to its powerful Fable 5 model: Axios
@financialjuiceTrump administration near to permit Anthropic to reinstate access to its advanced Fable 5 model - Axios
@axiosScoop: Powerful Anthropic model, Fable 5, on track to return soon
@axiosRT @mikeallen: 🚨NEW in @axios AM: The administration is close to greenlighting restoration of Claude Fable 5, the powerful Anthropic model…
@firstsquawkTrump administration near to let Anthropic regain access to its powerful Fable 5 model - Axios
@firstsquawkTrump administration near to permit Anthropic to reinstate access to its advanced Fable 5 model - Axios
@kimmonismusPer Axios: Fable 5 is expected to be back and available starting next week. Let’s hope it won’t be too heavily guardrailed or lobotomized, and that access will be broad.
@cto_juniorMy read on this: Since they haven’t updated series to GPT-6, it still seems to be same base as GPT-5.5 Secondly, the Terra model is mini. I expect it to have same shortcomings as all other mini models where it sucks at recognising user i
@reutersUS close to allowing Anthropic to restore Fable 5 model, Axios reports
@dnystedtRT @andrewdfeldman: Yesterday, @OpenAI announced GPT-5.6 Sol - their most powerful model ever. It launches on @Cerebras in July. Frontier…