Grok 4.5 - SpaceX AI's next-generation flagship large language model

What is Grok 4.5?

Grok 4.5 is SpaceX AI's (formerly xAI) next-generation flagship large language model, built on the 1.5 trillion parameter V9 architecture. During the supplemental training phase, the model deeply integrates Cursor programming data, significantly enhancing code generation and software engineering capabilities. Elon Musk positions it as an Opus 4.7-level model, emphasizing faster inference speed, higher token efficiency, and lower usage costs. In third-party benchmarks, Grok 4.5 ranks fourth globally on the GDPval-AA v2 benchmark and is already publicly available.

Main features of Grok 4.5

• Intelligent code generationBased on deep training of Cursor programming data, it supports code completion, bug fixing, project refactoring, and automated test generation.
• Multi-step engineering task processingCovers cross-file dependency analysis, architecture design, and technical documentation generation; proficient in long-term software engineering processes.
• Asynchronous execution of intelligent agentsIt supports long-running AI Agent tasks and can handle complex multi-round decision-making and office automation processes.
• Rapid reasoning outputWith a generation speed of 80 tokens per second and low response latency, it is suitable for real-time interactive scenarios.
• Multilingual programming supportIt performs well in benchmarks such as SWE-Bench Multilingual and supports cross-language code understanding and conversion.
• Platform native integrationIt is already embedded in the Grok Build dialog platform and Cursor IDE, so developers can use it without switching environments.
• Standard API AccessThe SpaceXAI console provides an OpenAI-compatible interface, supporting enterprise-level system integration and batch calls.

Technical principles of Grok 4.5

• Architecture ScaleBased on the brand-new 1.5 trillion parameter V9 architecture, it significantly improves model capacity and expressive power compared to its predecessor.
• Data EngineeringThe training data underwent rigorous deduplication, quality scoring, and domain filtering, and Cursor programming data was introduced as a supplementary training set to specifically enhance code understanding and generation capabilities.
• Training infrastructureIt employs tens of thousands of NVIDIA GB300 GPUs for large-scale training, along with stable operation technology to ensure uninterrupted training over long periods.
• Reinforcement learning optimizationIt covers hundreds of thousands of multi-step software engineering tasks and improves the inference efficiency of models in complex long-term tasks by supporting asynchronous training of agents that run for a long time.
• Reasoning efficiencyOptimizing decoding strategies and inference path planning, the SWE Bench Pro task consumes an average of only 15,954 tokens, about a quarter of the top models in the same category, achieving a dual compression of speed and cost.

How to use Grok 4.5

• Use via Grok BuildUsers can directly log in to the Grok Build platform and interact with the Grok 4.5 model through the chat interface, which is suitable for daily Q&A, code writing and document processing.
• Use via Cursor compilerGrok 4.5 has been integrated into the Cursor programming environment, allowing developers to call models within the IDE for code completion, bug fixing, and project refactoring, achieving seamless embedding of programming workflows.
• via API callDevelopers can log in to the SpaceXAI console to obtain the API key and access it according to the standard OpenAI compatible interface format.

The core advantages of Grok 4.5

• Performance comparable to the top tierInternal evaluation is roughly equivalent to Opus 4.7, and the Terminal-Bench 2.1 score is 83.3%, placing it in the top tier of large models.
• Ultimate cost advantageGDPval-AA v2 has a single task cost of only $0.49, which is nearly 90% cheaper than the top three models and lower than GLM-5.2 and Kimi K2.6.
• Token efficiency leadsSWE Bench Pro tasks consume an average of 15,954 tokens, which is 4.2 times less than Opus 4.8's 67,020 tokens, directly reducing users' actual expenditure.
• Extremely fast reasoning speedOutput speed reaches 80 tokens per second, meeting the standard for fast model operation and enabling more timely response to long-term tasks.
• Outstanding engineering capabilitiesIt performs excellently in software engineering benchmarks such as DeepSWE 1.0 and SWE-Bench Pro. After Cursor data enhancement, its code generation and multi-step engineering task processing capabilities are significantly improved.

Grok 4.5 project address

• Project official websitehttps://x.ai/news/grok-4-5

Grok 4.5 Comparison with Similar Products

Application scenarios of Grok 4.5

• Intelligent software developmentIt assists in code completion, bug diagnosis, project refactoring, and automated test generation in IDEs such as Cursor, making it suitable for full-stack development and large codebase maintenance.
• Multi-step engineering tasksIt handles complex, long-term software engineering processes, such as cross-file dependency analysis, architecture design, and automatic generation of technical documentation.
• Enterprise API Integration: Connect to internal systems through the SpaceXAI console to build low-cost, high-throughput customer service robots, data analysis, and business process automation tools.
• Terminal Programming AssistanceThe Grok Build platform enables the rapid generation of scripts, query commands, and configuration files, making it ideal for operations engineers and data scientists to improve their daily efficiency.