MeteoalarmOrange High-temperature Warning issued for Latvia (39 counties)Alerts
Saturday, 27 June 2026
Rīga TV

World and Latvian news in one place

TechnologyPublished: 27 June 2026 at 07:36

OpenAI begins limited preview of GPT-5.6 with three variants

OpenAI has launched a preview of its GPT-5.6 series, offering three variants—Sol, Terra, and Luna—to a select group of trusted partners, with a broader release expected in weeks.

Foto: Engadget

OpenAI has announced a preview of its GPT-5.6 model series, made available to a small group of trusted partners. The new family includes three variants: Sol, the most powerful model; Terra, designed for everyday use with performance similar to GPT-5.5 but at half the cost; and Luna, the lowest-cost option.

The company plans a wide release in the coming weeks. Prior to this, OpenAI gave the US government a preview of GPT-5.6 and its capabilities. At the administration's request, the model is being previewed to a small group of trusted partners whose participation has been shared with the government. OpenAI stated that it does not believe this government access process should become the long-term default, calling it a short-term step to allow for a public release soon.

Earlier this month, President Trump signed an AI cybersecurity order requiring companies to submit their most powerful models for voluntary government review 30 days before public release. According to a New York Times report, OpenAI, Anthropic, Google, xAI, and Microsoft had been giving early access to the government even before the order. Meta was the only holdout, and the government has been urging it to submit its models.

GPT-5.6 introduces a "max" reasoning effort, giving Sol more time to reason deeply. Sol is also OpenAI's most capable model for cybersecurity, helping find and fix vulnerabilities. OpenAI says Sol has strengthened protections for high-risk activities and sensitive requests. All variants include safeguards to withstand real-world adversarial pressure. Additionally, GPT-5.6 is trained to refuse "prohibited cyber assistance," including jailbreak attempts. The company spent 700,000 GPU hours identifying universal jailbreaks and developing countermeasures, and pledges a rapid-response process for newly discovered jailbreaks.

This focus likely stems from Anthropic's experience. A few weeks ago, Anthropic suspended access to its Mythos 5 and Fable 5 models after a government directive, reportedly due to concerns over jailbreaking. The company has since begun lifting access blocks, with US government permission to redeploy Mythos to select organizations.

Pricing: Sol at $5 per million input tokens and $30 per million output; Terra at $2.50 input and $15 output; Luna at $1 input and $6 output.

Comments

0/1500

Comments are automatically moderated. No hate, threats, personal data or spam.

Loading comments…

More in this category