xAI launches Grok 4.7 for coding & knowledge work
xAI has launched Grok 4.7, its new model for coding and knowledge work. The model is available through Cursor, the Grok API, and other software tools. Grok 4.7 was trained with a longer reinforcement learning run on harder tasks, focusing on problems that can take many hours to complete. xAI benchmarked Grok 4.7 against rivals like GPT-5.6 Sol and Fable 5.1 Max, showing mixed results across various professional tasks.
xAI's Grok 4.7 launch shows the escalating competition for developer workflows. While Grok 4.7 improved over its predecessor, Grok 4.6, in all reported benchmarks, it did not consistently lead the field. For instance, Grok 4.7 scored 46.3% on CursorBench 4.0 for long coding tasks, trailing Fable 5.1 Max at 51.8%. This reflects a market where vendors increasingly trade off cost, speed, and safeguards against peak performance in specific niches.
The model's pricing, USD $2 per million input tokens for the standard variant, positions it for direct competition in Asia's growing developer market. Companies like Alibaba Cloud and Tencent Cloud, which offer their own AI models and developer platforms, will face renewed pressure to match or exceed Grok 4.7's performance in specialized tasks like electrical engineering, where Grok 4.7 scored 64.0% against GPT-5.6 Sol's 39.4%. This pushes Asian cloud providers to enhance their own developer tools and model capabilities.
The move to offer invite-only access to Grok 4.7 red-team functions for cybersecurity partners is a key development. This suggests xAI aims to penetrate the security testing market. The test for Asian cybersecurity firms will be whether Grok 4.7's 3.3% rate of allowing risky dual-use prompts on HackerBench v0.3 proves superior to existing tools. This could drive adoption among Asia's large cybersecurity sector.
Related reading
6 stories
Huawei Ascend 960 SuperPoD aims to surpass Nvidia AI hardware in China — 960E with over 4,000 NPUs and an upgraded UnifiedBus Interconnect aims for 10 trillion parameters
The escalating competition for AI developer workflows also drives hardware innovation, as seen with Huawei's Ascend 960 SuperPoD.
Nearly Half of Singaporeans Are Comfortable Letting AI Apply for Credit
Deals in brief: Singapore-based dtcpay closes Series A round, 01F Group invests in Dana, ten China investments, and more
China eyes chip-industry foothold with a tiny Han Xin code to challenge US standard
LexisNexis Launches AI Tool That Adapts to Changing Fraud Patterns
