OpenAI introduced its newest mannequin household on Thursday, introducing a strong set of packages into the more and more crowded discipline of AI merchandise.
GPT-5.6 is available in three variants: Sol (thought of the flagship), Terra (a extra intermediate possibility), and Luna (a budget-friendly possibility). These fashions develop what customers can do in quite a lot of fields and promise highly effective capabilities in company work, coding, and even scientific analysis.
CEO Sam Altman guarantees that his firm’s newest mannequin is orders of magnitude extra environment friendly and cost-effective than earlier variations, not too long ago telling CNBC that Sol is 54% extra token environment friendly in terms of AI coding duties.
Most notably, the corporate calls 5.6 “probably the most highly effective cybersecurity mannequin thus far that achieves frontier efficiency with considerably fewer tokens.”
Certainly, there was loads of noise concerning the mannequin’s cyber capabilities, with the Trump administration beforehand looking for to restrict its deployment, ostensibly as a result of fears that it might be misused. GPT-5.6 helps defensive actions comparable to menace modeling, code overview and patching, and blue teaming (simulating assaults by yourself techniques to search out weaknesses earlier than an actual hacker does).
OpenAI additionally launched a brand new instrument known as ChatGPT Work. Because the identify suggests, this instrument is designed as a office companion for company groups, operating on desktop, internet, and cell to assist with on a regular basis workplace duties comparable to creating paperwork, spreadsheets, and shows.
OpenAI’s newly introduced household of fashions follows comparable releases from opponents SpaceXAI and Meta this week.
Nonetheless, GPT-5.6 and its accompanying advertising look like designed to focus on OpenAI’s main adversary, Anthropic. Anthropic has managed to turn out to be a likable underdog within the AI race by specializing in enterprise prospects and gaining share of assist in consequence.
To not be outdone, OpenAI cites a notable benchmark metric, the Synthetic Analytics Coding Agent Index, and claims that its newest household of fashions outperforms Anthropic’s fashions in each manner.
OpenAI calls Sol “one of the best coding mannequin ever” and explicitly compares it to Anthropic’s not too long ago launched (and much-hyped) Fable. Utilizing the Coding Agent Index, OpenAI claims that Sol “units a brand new bar at 80, 2.8 factors above Fable 5, whereas utilizing lower than half the output tokens, taking lower than half the time, and costing a few third much less.”
Moreover, “the advantages prolong throughout the household, with Terra barely outperforming Fable 5 and Luna outperforming Opus 4.8.”
The corporate says 5.6 is now obtainable for ChatGPT, Codex, and OpenAI APIs. The provision worth per million tokens will probably be: Sol is $5 enter / $30 output, Terra is $2.50 enter / $15 output, and Luna is $1 enter / $6 output.
In the event you purchase by means of hyperlinks in our articles, we might earn a small fee. This doesn’t have an effect on editorial independence.
