SpaceXAI releases Grok 4.6, improving performance and challenging top AI models
SpaceXAI, Elon Musk's artificial intelligence company, has launched its latest AI model, Grok 4.6. The new model shows significant performance gains and is positioned as a competitive option among leading AI systems.
SpaceXAI, formerly known as xAI, has released Grok 4.6, its newest artificial intelligence model. The model focuses on enhancing capabilities for long-running agents, coding tasks, and knowledge work, aiming for cost-efficiency in these areas.
The model scored 61 on the third-party Artificial Analysis Intelligence Index, surpassing China's Moonshot Kimi K3 and matching OpenAI's GPT-5.6 Sol Max. This represents a five-point improvement over its predecessor, Grok 4.5 High. Anthropic's Claude Opus 5 and Fable 5 remain at the top of the index.
Grok 4.6 demonstrates substantial improvements over Grok 4.5 in benchmarks for coding, terminal interaction, knowledge work, and agent performance. Its API pricing, starting at $2 per million input tokens and $6 per million output tokens, positions it competitively within the market for both proprietary and open-source models.
SpaceXAI states that Grok 4.6 is designed to maintain focus and perform consistently over extended task sequences, including research, analysis, and application development. The company highlights that the model underwent an extended supplemental training period and received enhancements in machine learning, particularly focusing on agentic behavior.
Performance tests show Grok 4.6 achieving notable scores in benchmarks such as APEX-Agents and CursorBench. While not leading in every category, the release strengthens SpaceXAI's position in the competitive landscape of advanced AI development.