xAI Grok 4.5 Publicly Launched, Claims Opus-Class Performance

On July 9, xAI opened the Grok 4.5 model to the public. The official description claims its performance reaches the Opus class. This is the first time since the Fable 5 export control ban took effect on June 12 that all major frontier AI labs simultaneously have publicly available models. On the same day, OpenAI released GPT-5.6 Sol, Terra, and Luna to ChatGPT users and API developers, while Gemini 3.5 Pro remains in preview.

Fact Restoration

According to public information, xAI completed the public release of Grok 4.5 on July 9, emphasizing that its performance is comparable to the Opus class. OpenAI concurrently ended its 13-day government-coordinated preview, fully opening three GPT-5.6 models. Prior to this, the export controls implemented on June 12 had restricted the availability of some models, with July 9 marking the point at which this restriction was broken. Gemini 3.5 Pro remains the only expected model still in preview.

Mechanism Breakdown

The reason multiple models were publicly released on the same day is the end of the regulatory window. After the June 12 ban was enacted, each lab entered a limited preview phase. On July 9, OpenAI ended its preview with approximately 20 partners, while xAI directly opened Grok 4.5. The combination of these two events created a simultaneous launch. xAI chose to release on this day, achieving synchronization with OpenAI and fulfilling the condition that "every frontier lab has a publicly available model."

Industry Impact

For developers, the public release of Grok 4.5 means an additional directly callable option, and its claim of Opus-class performance provides the potential for direct comparison with existing models. For enterprise users, having multiple models available simultaneously reduces the risk of relying on a single source, allowing users to choose between Grok 4.5 and the GPT-5.6 series based on actual workloads. In terms of competitive landscape, this is the first time that models from three labs are concurrently online, while Gemini 3.5 Pro's preview status places it in an observational position for now.

Strategic Judgment

Analysis shows that the most likely scenario going forward is that each lab will continue to expand model coverage through public channels to maintain simultaneous online availability. xAI may further clarify Grok 4.5's applicability in specific scenarios such as coding, while OpenAI may adjust access tiers for the GPT-5.6 series. If Gemini 3.5 Pro exits preview, it will again shift the balance of concurrent availability. Overall, the configuration formed on July 9 provides a baseline for subsequent version iterations.

The above content is strictly based on confirmed release facts and does not convert preview-phase information into definitive conclusions.