xAI makes Grok 4.7 available with a 500k-token API context

xAI's developer release notes list Grok 4.7 as available on the public xAI API. The model accepts text and images, supports a 500k-token context window, offers four reasoning-effort settings, and uses a two-tier price schedule based on prompt length. GitHub's changelog separately lists Grok 4.7 as available in GitHub Copilot, giving the release a second distribution path for developers.

Grok 4.7 targets long coding and agent workflows

xAI positions Grok 4.7 for coding, agentic tasks, and knowledge work. The documented API shape is text-and-image input with text-only output, a 500k context window, and no stated text-output limit. Reasoning effort can be set to low, medium, high, or xhigh; high is the default.

That context size is the practical headline for teams building code agents, repository analysis, or long-running tool workflows. It leaves room for source files, tool results, and prior reasoning in a single request, although a larger window does not guarantee lower cost or better answers. Applications still need truncation, retrieval, and tool-result controls when inputs grow beyond useful signal.

xAI publishes a prompt-length price split

For prompts below 200,000 tokens, xAI lists pricing of $2 per million input tokens, $0.50 per million cached input tokens, and $6 per million output tokens. Above that prompt threshold, the listed rates rise to $4, $1, and $12 respectively.

The qualifier matters for long-context workloads. A system that stays below the threshold can have a materially different request cost from one that sends a large repository snapshot or accumulated agent transcript. Teams should model both uncached and cached input, then measure prompt sizes at the production percentile rather than budgeting from the short-context rate.

Access is split across the xAI API, Cursor, and Copilot

The public API exposes grok-4.7. xAI also documents Grok 4.7 Fast, described as the same model at twice the token rates, as available through Cursor and Grok Build rather than the public xAI API. Those are separate access paths with separate billing and product controls.

GitHub's changelog lists Grok 4.7 in GitHub Copilot as well. Copilot availability does not establish identical API limits, model behavior, or billing terms, so organizations should verify the model picker and plan entitlement in their own GitHub account before changing a default model.

What developers should verify before routing traffic

• Confirm that grok-4.7 is enabled for the intended xAI API account and region.
• Recalculate costs at the 200,000-token prompt boundary, including cached input.
• Test medium and high reasoning effort on representative coding and tool-use tasks.
• Compare API, Cursor, Grok Build, and Copilot access separately; the product names do not imply shared quotas.
• Keep a fallback route while availability and pricing settle across integrations.

Grok 4.7 is now a concrete API option for developers who need long context and configurable reasoning. The next operational checkpoint is usage data from real workloads: whether the larger context reduces retrieval overhead enough to justify the higher long-prompt tier.

Sources and methodology

The xAI Developer Release Notes are the primary source for model availability, context, capabilities, reasoning settings, and pricing. GitHub's independent Changelog is used only to corroborate availability in GitHub Copilot. Model directories, social posts, and Replicate or fal listings were treated as discovery signals rather than proof of release status.