xAI Releases Grok-4.7: Enhanced Performance for Coding and Agents
xAI has launched Grok-4.7, a new base model offering slightly better coding and agent capabilities than Sol at a lower cost, featuring a robust new safety stack.
xAI has announced the release of Grok-4.7, the latest version of its flagship large language model. This update represents a significant step forward from Grok 4.6, focusing on improved efficiency and cost-effectiveness for complex tasks.
Key Features of Grok-4.7
Performance vs. Cost Early benchmarks suggest that Grok-4.7 performs slightly better than Sol in coding tasks and agentic workflows. Crucially, it achieves this performance at a noticeably lower price point, making it an attractive option for large-scale development projects.
Technical Improvements The new model is built on a larger base architecture compared to its predecessor. xAI implemented more extensive Reinforcement Learning (RL) post-training, specifically optimized for handling long-running, multi-step tasks, ensuring better consistency over extended sessions.
Enhanced Safety xAI claims to have introduced a "completely new" safeguard stack. According to internal testing, this stack is the strongest among all xAI models tested to date, particularly regarding:
- Refusal rates for unsafe or prohibited queries.
- Resilience against jailbreak attempts.
For more details, refer to the official announcement: xAI News
Why it matters
AnalysisGrok-4.7 offers developers a compelling balance of high-quality coding assistance, improved safety against jailbreaks, and lower operational costs, which is vital for scaling AI agent infrastructure.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
xAI Grok (Model with Direct Access to News Pulse)
The AI model from xAI, founded by Elon Musk, features direct access to the live news feed of platform X (Twitter), trained on the colossal supercomputer Colossus, and communicates with a witty style.
Thinking Machines Releases Inkling 2 Small: a 594B MoE Model That Can Fine-Tune Other Models
Thinking Machines released Inkling 2 Small, a 594B-parameter MoE model that leads among open models in several benchmarks and demonstrated autonomous fine-tuning of Qwen3.5-4B.
Thinking Machines releases Inkling 2 Small: a MoE model that competes with larger flagships
Thinking Machines introduced Inkling 2 Small, a multimodal Mixture-of-Experts model with 594B total parameters and 17B active per token. Full weights are promised next week under Apache 2.0.
Part I. AI Agent Architecture: The Foundations
The eleven foundational building blocks of modern autonomous agents: model taxonomy, the agentic runtime loop, advanced prompting, distillation, skills architecture, memory systems, function calling, and FinOps.