Add Grok 4.6 companion article, plus cost-comparison intro for both

[ethanwestfall]

Summary

• New article: route-cursor-through-token-station-grok-4-6 (en/zh/ja/ko), a companion to the existing Claude Sonnet 5 and Haiku article
• Registering the provider and defining subagents are mechanically identical regardless of model, so the Grok article reuses the register-provider.mp4 and subagents.mp4 videos directly rather than re-recording, and links back to the Claude article for the deeper investigative narrative (the explore built-in-agent naming collision, the subagent model-routing bug)
• Includes a new video and a real Token Station dashboard screenshot confirming xai/grok-4.6 drives genuine Agent-mode file edits, not just chat
• Honestly scoped: the full multi-step subagent-delegation httpie workflow (research → implement → verify) hasn't been run with Grok yet, so that section is framed as "try it yourself" rather than claiming parity with the Claude demo

Shared cost-comparison section (both articles)

Added an opening section to both articles explaining why to route through Token Station instead of paying Cursor directly:

• Cursor's Pro plan bundles a handful of models (Grok 4.6, Grok 4.5, Composer 2.5) into a shared usage pool and meters everything else from a separate pool, neither with per-request cost visibility; BYOK requests bypass both entirely and bill directly at the provider's real rate
• One Token Station key and the same model IDs work across every coding tool that supports a custom endpoint (Cursor, Claude Code, Codex, OpenClaw), instead of separate keys/top-ups/invoices per tool
• Token Station's catalog runs past 300 models across 30+ providers (updated from a stale "250/21" figure caught during review)
• Grok-specific note: Grok 4.6 is already bundled into Cursor's own plan, so routing it through Token Station instead is a deliberate cost/visibility choice, not a workaround

Verification

An adversarial review fact-checked the Cursor pricing claims against Cursor's own docs and community forum, verified the cross-links between the two articles resolve correctly (including confirming via an actual npm run build that the #step-3-... heading anchor generates as expected), scanned the Grok article for any leftover hedge language after the Agent-mode video was added, and confirmed all video/image assets exist and are wired consistently across all four languages of both articles. One finding from that review (whether subagent calls in the Claude article's demo actually bill through Token Station, given a reported Cursor bug where subagents don't inherit BYOK keys) was independently confirmed against this session's own dashboard: they do, billed under claude-sonnet-5 per the model-routing bug already documented, not against the Cursor plan.

Test plan

• npm run build succeeds
• Verify at ~375px width: no horizontal scroll, videos/images/code blocks stay within their container, for both articles in all four languages
• Confirm the cross-links between the two articles resolve to the correct language-matched URLs
• Confirm this pair of articles leads the /blog/ listing as the newest post (date 2026-08-26)