xAI has released Grok 4.7 for coding, agentic tasks and knowledge work with a 500,000-token context window, published API pricing and a gradual rollout across paid GitHub Copilot plans.
RELATED BUYER PROFILEReview Grok 4.7 pricing, access and independent-test-pending evidencePricing · access · strengths · limitations →RELATED DECISION GUIDERun a same-work model and agent pilotWorkflow · evidence · risk · governance →CONTINUE THE DECISIONBuild a controlled AI stack for a software teamEvidence · workflow · next action →CONTINUE THE DECISIONMeasure complete cost per accepted taskEvidence · workflow · next action →CONTINUE THE DECISIONReview the separate GitHub Copilot buyer profileEvidence · workflow · next action →CONTINUE THE DECISIONAdd Grok 4.7 to an existing shortlistEvidence · workflow · next action →What you need to know
- The public API model ID is grok-4.7, with a 500,000-token context window, text and image input, text output and four reasoning-effort levels
- Standard API pricing is $2 per million input tokens and $6 per million output tokens; the faster route costs more and is not available through the public xAI API
- GitHub Copilot access is rolling out gradually across Pro, Pro+, Max, Business and Enterprise, with organisation administrators able to manage model access
The verified launch boundary
Short answer: a real release, not an independent performance result
xAI released Grok 4.7 on 21 September 2026 for coding, agentic tasks and knowledge work. The model is available through the xAI API, Cursor, Grok Build, model routers and other coding harnesses. GitHub separately announced a gradual rollout in Copilot. The model code, context window, price and product routes are verified. xAI's speed, safety and benchmark comparisons remain provider-reported claims until they are reproduced on independent and buyer-specific work.
What is actually new
xAI describes a larger base model than Grok 4.6, a longer reinforcement-learning run on harder multi-hour tasks, stronger self-verification and better long-context management. The API accepts text and image input and produces text output. Reasoning effort can be set to low, medium, high or xhigh. Supported tools include function calling, web search, X search and code execution. These capabilities make the model a plausible candidate for repository work and longer professional tasks, but they do not establish accuracy on a particular codebase, document set or regulated workflow.
API pricing starts at $2 input and $6 output
The standard public xAI API rate is $2 per million input tokens and $6 per million output tokens. A US regional endpoint carries a 10% token-price premium. xAI also describes a Fast variant at twice the standard token rates, but its documentation says that route is limited to Cursor and Grok Build and is not available on the public xAI API. Grok Build has a free starting route; that does not make API usage or the complete production workflow free.
A 500K window still needs cost and context discipline
The API documentation lists a 500,000-token context window and recommends a prompt cache key so related requests reach the same server. Without reliable cache hits, repeated long-context work can be billed at the full input rate. Long agent loops can also require context compaction. Buyers should record actual billed tokens, cache-hit behaviour, tool calls, retries and review time rather than multiplying the headline rate by an optimistic prompt estimate.
GitHub Copilot is a separate commercial route
GitHub says Grok 4.7 is rolling out across Copilot Pro, Pro+, Max, Business and Enterprise, including Visual Studio Code, Visual Studio, Copilot CLI, the cloud agent, the Copilot app, JetBrains, Xcode and Eclipse. GitHub describes provider-list usage billing, while plan entitlements and organisation policy still apply. Business and Enterprise administrators can manage the model in Copilot settings. A model appearing in a picker does not mean every account has it immediately or that xAI API pricing is the team's final Copilot bill.
The benchmark table is a test hypothesis
xAI publishes results across CursorBench, DeepSWE, Terminal-Bench, professional-work, legal and clinical-reasoning tests. It also claims stronger refusal and jailbreak resistance. These figures are useful for selecting pilot tasks, but they are not an independent HubAI measurement, a safety certification or proof of suitability for legal, clinical or security decisions. Any high-consequence output still requires qualified review and workflow-specific controls.
Run one same-work evaluation across the routes
Choose five representative tasks with frozen inputs and acceptance tests. Run the same tasks through Grok 4.7 in the intended route and one existing model. Keep repository permissions, tools, reasoning effort and context fixed. Measure accepted-task rate, unsupported changes, tests passed, human review minutes, tool failures, cache hits, latency and total billed cost. If GitHub Copilot is the intended purchase, measure Copilot consumption and policy behaviour rather than substituting xAI API estimates.
HubAI buyer verdict
Grok 4.7 belongs on a coding and agent shortlist because the model identifier, public API rate, long-context limit and multiple deployment routes are concrete. No numerical HubAI Score is assigned before an independent same-work test. The sensible starting point is standard speed and high reasoning on a bounded task; move to xhigh, Fast or a wider agent loop only when the extra spend improves accepted work after review.
Evidence limits and editorial disclosure
Independent editorial coverage; not sponsored. Model architecture descriptions, benchmark results, safety measurements, pricing and availability are attributed to xAI, GitHub and their official documentation, checked on 22 September 2026. HubAI has not independently tested Grok 4.7. The visible UK Google Trends first 25 contained no direct Grok or Grok 4.7 query during this review, so no search-volume or breakout claim is made. The cover is an original conceptual illustration, not a product interface, provider logo or benchmark chart.
HUBAI VIEWTreat the release as a same-work pilot, not a benchmark verdict. Compare accepted results, review time, cache behaviour, tool failures and complete cost across the exact route your team will buy.
Buyer decision signal: New frontier model · API and Copilot rollout
What to verify next
1Confirm which route is being bought: xAI API, Cursor, Grok Build, a model gateway or GitHub Copilot
2Freeze representative tasks, permissions, tools, reasoning effort and acceptance tests
3Verify the actual model identifier, account access and regional endpoint
4Record input, cached input, output, tool, retry and infrastructure charges
5Measure accepted results, unsupported changes, tests and human review minutes
6Keep high-consequence legal, clinical, security and financial outputs under qualified review
7Retain an alternative model, rollback path and administrator control over access
Read the evidence
Capabilities, availability and prices can change. HubAI keeps analysis separate from the underlying official material.
01xAI: Introducing Grok 4.7 — 21 September 2026Open source ↗02xAI developer documentation: Grok 4.7 model, pricing and limitsOpen source ↗03GitHub: Grok 4.7 rollout in Copilot — 21 September 2026Open source ↗04GitHub documentation: supported Copilot models and plansOpen source ↗
Models
Models
Models
Models