23 hrs ago
SpaceX AI Unveils Faster, Cheaper Grok 4.7 Coding Model
SpaceX AI says it has released a new computer program called Grok 4.7.
It is designed to help people write code and complete other difficult knowledge tasks.
The program can work on problems for longer and check some of its own work.
It is also meant to create documents and presentations more effectively.
The company says it has added stronger safety protections.
These protections are intended to help the model resist attempts to bypass its rules.
Grok 4.7 can be used for legitimate cybersecurity work while limiting some dangerous requests.
It is available on several coding platforms and costs $2 for every million input tokens and $6 for every million output tokens.
SpaceX AI introduced Grok 4.7 as its most capable model for coding and knowledge work.
The model is trained for longer tasks, can check its own progress, and handles larger contexts.
Grok 4.7 reportedly improves document, presentation, conversational, and general knowledge-work capabilities.
New safeguards target jailbreak resistance and dual-use areas such as cybersecurity and biological work.
The model is available through Cursor, Grok Build, APIs, coding tools, model routers, and cloud platforms at $2 per million input tokens and $6 per million output tokens.
- Who
- SpaceX AI, also referred to as xAI in the article, released Grok 4.7.
- What
- The company unveiled a coding and knowledge-work model with longer reasoning, self-checking, cybersecurity features, and stronger safeguards.
- Where
- It is available through Cursor, Grok Build, the Grok API, third-party coding harnesses, model routers, and cloud platforms.
- When
- The article states that Grok 4.7 is available now.
- Why
- The model is intended to improve extended coding and knowledge-work tasks while providing stronger safety and cybersecurity protections.
Key facts
- Model
- Grok 4.7
- Primary uses
- Coding and knowledge work
- Input price
- $2 per million input tokens
- Output price
- $6 per million output tokens
- Availability
- Cursor, Grok Build, the Grok API, third-party coding harnesses, model routers, and cloud platforms
- Cybersecurity benchmark
- The article says 3.3% of potentially dangerous dual-use prompts were restricted on HackerBench v0.3
- Faster variant
- A version with twice the output speed is offered at a higher price
Sources
Related news
OpenAI Launches GPT-6 Astra, Claims Edge Over Claude
OpenAI Rolls Out GPT-6 Astra With Cybersecurity Guardrails
Anthropic launches Claude Fable and Mythos 5.1 models
OpenAI's GPT-6 Astra Promises Agentic Power, Raises Cybersecurity Concerns
OpenAI launches GPT-6 Astra for autonomous computer tasks
Anthropic Launches Fable 5.1 and Mythos 5.1
Grok Bot reaches 418,000 weekly users one month after launch
US Department of War Expands GenAI.mil With ChatGPT, Grok
AI Model Releases Outpace Independent Testing, Raising Evaluation Concerns
SpaceXAI unveils Grok 4.6, claims frontier-level performance at lower cost
