Grok 4.7 Benchmarks: How xAI's New Model Stacks Up Against Claude Fable 5.1 and GPT-6

On September 21, 2026, xAI launched Grok 4.7, its latest model for coding and knowledge work, maintaining its predecessor's competitive pricing at $2 per million input tokens and $6 per million output tokens. While Grok 4.7 trails Claude Fable 5.1 and GPT-6 on the Artificial Analysis Intelligence Index, it leads its class in specific coding benchmarks and excels in legal-agent and electrical-engineering tasks.

Grok 4.7 Performance Overview

The Artificial Analysis Intelligence Index provides a broad measure of AI model capabilities. On this index, Grok 4.7 achieved a score of 46. In comparison, both Claude Fable 5.1 and GPT-6 recorded higher scores of 53, indicating that Grok 4.7 trails these rivals in overall intelligence as measured by this specific benchmark.

Coding and Engineering Strengths

Despite the overall intelligence gap, Grok 4.7 demonstrates notable strengths in specialized areas. It leads its class on CursorBench 4.0, scoring 46.3%, which suggests strong performance in coding-related tasks. Furthermore, xAI's new model surpasses GPT-5.6 Sol Max and Fable 5.1 Max in specific legal-agent and electrical-engineering tasks, highlighting its targeted capabilities in these professional domains.

Agentic Task Performance

In the realm of agentic tasks, Grok 4.7's performance is more varied. On Terminal-Bench 4.0, a benchmark for agentic capabilities, Grok 4.7 scored 38.0%. This places it significantly behind Fable 5.1 Max, which achieved 57.9% on the same benchmark. This disparity suggests that while Grok 4.7 excels in certain specialized applications, its broader agentic reasoning capabilities may not yet match those of its top competitors.

Pricing and Accessibility

xAI has maintained a consistent pricing strategy for Grok 4.7, keeping it at $2 per million input tokens and $6 per million output tokens. This matches the pricing of Grok 4.6, potentially offering a cost-effective option for developers and businesses. Grok 4.7 is accessible through the Grok API, Cursor, and Grok Build, providing multiple avenues for integration and use.

Enhanced Safeguards and Resistance

According to xAI, Grok 4.7 incorporates a new safeguard stack, making it the company's most robust model yet in terms of refusal capabilities and jailbreak resistance. This focus on safety and ethical AI use is a critical development, addressing growing concerns about model misuse and harmful outputs.

Key Takeaways

• Grok 4.7 launched on September 21, 2026, with a new, larger base model.
• It is priced at $2 per million input tokens and $6 per million output tokens, matching Grok 4.6.
• Grok 4.7 scored 46 on the Artificial Analysis Intelligence Index, while Claude Fable 5.1 and GPT-6 scored 53.
• The model leads its class on CursorBench 4.0 at 46.3% for coding tasks.
• Grok 4.7 trails Fable 5.1 Max on Terminal-Bench 4.0 for agentic tasks.
• xAI states Grok 4.7 has enhanced safeguards and jailbreak resistance.

Conclusion

xAI's release of Grok 4.7 marks a continued effort to compete in the rapidly evolving AI landscape. While it demonstrates strong performance in specific areas like coding and certain engineering tasks, and maintains a competitive pricing model, it currently lags behind frontier models such as Claude Fable 5.1 and GPT-6 in broader intelligence and some agentic benchmarks. Its enhanced safety features are a notable improvement. Developers and organizations considering Grok 4.7 should evaluate its specialized strengths against their specific application needs, particularly for coding and knowledge work where it shows promise.

Sources

• Introducing Grok 4.7 | SpaceXAI
• xAI is shipping Grok 4.6 and 4.7 back to back in a release...
• Grok 4.7 (xhigh) - Intelligence, Performance & Price Analysis | Artificial Analysis
• xAI launches Grok 4.7 at bargain prices, but benchmarks reveal a wide gap to Claude and GPT-6
• Grok 4.6 gets Aug. 7 launch date, Grok 4.7 follows