AI News (2026/5/4): xAI Launches Latest Flagship Inference Model Grok 4.3
Executive Summary:
xAI has launched the Grok 4.3 model, positioned as a pragmatic transitional version. The model's API prices have been reduced by 40%-60%, with an output speed of 196 Tokens/s and support for 1 million Token long context. It has shown significant improvements in agent tasks and office assistance, particularly in generating documents, tables, and presentations. However, the model lags behind GPT-5.5 and Claude Opus 4.7 in the Intelligence Index, with insufficient stability in complex reasoning and fact-checking, and an increased hallucination rate.
News Details
xAI launched its latest flagship inference model Grok 4.3 on Sunday, May 3rd. This model is positioned as a pragmatic transitional version, aimed at providing a more cost-effective choice for developers and businesses. Grok 4.3 has been optimized in several areas, particularly in agent tasks and office assistance capabilities. Despite these improvements, the model still lags behind GPT-5.5 and Claude Opus 4.7 in the Intelligence Index.
Key Points
API Pricing: The API price of Grok 4.3 has been reduced by 40%-60% compared to the previous version, significantly lowering the usage cost for developers.
Output Speed: The output speed of Grok 4.3 has reached 196 Tokens/s, an improvement of about 20% over the previous version. This enhancement makes the model more efficient in real-time applications.
Long Context Support: Grok 4.3 supports 1 million Token long context, enabling it to handle more complex text generation tasks such as generating long documents, tables, and presentations.
Agent Tasks and Office Assistance Capabilities: Grok 4.3 has shown significant improvements in agent tasks and office assistance capabilities. It performs particularly well in generating documents, tables, and presentations.
Intelligence Index Performance: Although Grok 4.3 has been optimized in multiple areas, its performance in the Intelligence Index still lags behind GPT-5.5 and Claude Opus 4.7. Specifically, it lacks stability in complex reasoning and fact-checking, and has seen an increase in the hallucination rate.
AI-ALL In-Depth Analysis
The launch of Grok 4.3 demonstrates xAI's efforts to balance performance and cost. By significantly reducing API prices and improving output speed, the model offers a more attractive option for developers. The support for 1 million Token long context makes it excel in handling complex text generation tasks, especially in office scenarios where its application potential is significant. However, the limitations of Grok 4.3 in complex reasoning and fact-checking also highlight its shortcomings in certain high-end application scenarios. For tasks requiring high accuracy and stability, developers may still need to consider other more mature models. Overall, Grok 4.3 is a pragmatic choice for the mid-range market, contributing to the popularization and application of AI technology.
