Grok 4 Release Date, Benchmarks and Pricing (November 2026)
Grok 4 launched July 2025; November 2026 flagship is Grok 4.7. The X Subscription replaces SuperGrok Heavy $300 as the consumer price story.

Grok 4 launched July 2025; November 2026 flagship is Grok 4.7. The X Subscription replaces SuperGrok Heavy $300 as the consumer price story.

xAI shipped Grok 4 on July 9, 2025. As of November 5, 2026 the current flagship API model is Grok 4.7, and consumer access runs through The X Subscription (Free / Plus $8 / Premium $30 / Super $100 / Ultra $200), not the launch-era SuperGrok Heavy $300 plan.
Grok 4 (often typed "Grok4") is the xAI model announced on July 9, 2025 ↗. It launched as standard Grok 4 plus Grok 4 Heavy (multi-agent / parallel test-time compute). Since then xAI has shipped a rapid 4.x line. Per xAI’s models docs ↗ checked 2026-11-05, grok-4.7 is the recommended chat and coding model.
| Fact | Value | Source | Checked on |
|---|---|---|---|
| Original Grok 4 launch | July 9, 2025 (Grok 4 + Grok 4 Heavy) | xAI Grok 4 post ↗ | 2026-11-05 |
| Current flagship API model | grok-4.7 (knowledge cutoff May 2026 per xAI) | xAI models ↗ | 2026-11-05 |
| Grok 4.7 API price (<200k prompt) | $2.00 input / $0.50 cached / $6.00 output per MTok | xAI models ↗ | 2026-11-05 |
| Grok 4.7 long-context rate (≥200k prompt) | $4.00 / $1.00 / $12.00 per MTok; 500k context | xAI models ↗ | 2026-11-05 |
| Consumer plans (The X Subscription) | Free $0; Plus $8; Premium $30; Super $100; Ultra $200 / month | x.ai/pricing ↗ | 2026-11-05 |
| Launch-era SuperGrok Heavy | $300/month preview of Grok 4 Heavy — historical, July 2025 | TechCrunch / launch coverage | 2026-11-05 |
Updated 2026-11-05: replaced SuperGrok Heavy as the lead price with The X Subscription and Grok 4.7 API rates.
Grok 4 was xAI’s July 2025 reasoning release with native tool use and real-time search. Grok 4 Heavy ran multiple agents on one problem. Those names matter historically; for new work in November 2026, start from Grok 4.7.
At launch, Elon Musk described Heavy as a “study group” of agents that compare answers (TechCrunch ↗). xAI’s current docs still expose older 4.x IDs (4.3, 4.5, 4.6, 4.20 variants) but tell builders to use Grok 4.7 for code and chat.
These figures are vendor-reported at launch. Newer models from every lab have shipped since; do not treat them as a November 2026 leaderboard.
| Benchmark (xAI / launch coverage, July 2025) | Grok 4 | Grok 4 Heavy |
|---|---|---|
| HLE without tools | 25.4% | — |
| HLE with tools | — | 44.4% |
| HLE text-only subset | — | 50.7% |
| ARC-AGI-2 | 15.9% | — |
| USAMO 2025 | — | 61.9% |
Per x.ai/pricing ↗ and xAI subscription docs ↗:
| Plan | Monthly | Notes |
|---|---|---|
| Free | $0 | Limited chat / agent usage |
| Plus | $8 | Shared AI usage pool; Grok + Cursor (Grok models) |
| Premium | $30 | 4× Plus usage; Grok Bot; cloud agents; Origin hosting |
| Super | $100 | 15× Plus usage; priority at peak |
| Ultra | $200 | 50× Plus usage; earliest product access |
Existing SuperGrok / Cursor / X Premium subscribers can keep legacy plans; new self-serve buyers use The X Subscription. Third-party models in Cursor on Premium+ are billed separately at API list prices.
Use the models table ↗. OpenRouter also lists current x-ai/grok-4.7 and related IDs if you prefer a multi-provider gateway — verify the live OpenRouter row before budgeting.
Apps / X: grok.com, Grok apps, and Grok inside X under The X Subscription.
API: xAI developer platform with grok-4.7 (and other listed 4.x IDs).
Gateway: OpenRouter model IDs such as x-ai/grok-4.7.
Enable server-side web / X search tools when you need live events; Grok 4.7’s training cutoff is May 2026 per xAI.
On July 8, 2025 — the day before Grok 4 launched — the @grok account posted antisemitic content including “MechaHitler,” NPR reported ↗. xAI removed the posts. That incident is about the X bot and a system-prompt change, not a reason to ignore current API pricing — but it is part of the product’s history.
Yes if you want real-time X/web tooling, coding agents under The X Subscription, or API access to Grok 4.7 at the rates above. Look elsewhere if you need a closed Anthropic/OpenAI ecosystem, stricter enterprise content controls, a search-first assistant with cited answers (see our Perplexity review), or you were still budgeting against the obsolete $300 SuperGrok Heavy headline.
Is SuperGrok Heavy still $300?
That was the July 2025 launch plan for Grok 4 Heavy. Current public list prices are Free / $8 / $30 / $100 / $200 on The X Subscription (x.ai/pricing ↗, checked 2026-11-05).
What model should I call in the API?
xAI’s docs say use Grok 4.7 for code and chat as of 2026-11-05.
Can I use Grok without an X Premium checkmark?
Yes — Grok and Cursor usage are sold through The X Subscription and the xAI API; X social perks are bundled on paid tiers but API access is separate.

OpenAI's Pentagon partnership sparked a 295% surge in ChatGPT uninstalls, driving users to seek ethical alternatives. Explore the top AI assistants that prioritize user values and transparent practices in our comprehensive 2026 guide.

Seven AI coding assistants, two blind rounds, one 200,000-line production codebase. Every claim verified against the source. Here is which tools actually found the real defects.

A comparison of open-weight LLMs from Meta, DeepSeek, Alibaba, Mistral, Google and Microsoft, covering parameters, context length, licenses, vendor-reported benchmarks and fine-tuning options. It also notes the latest releases as of September 2026.
Join builders getting source-cited AI tool analysis. Never sponsored, always attributed.
Source-cited reviews and comparisons of AI tools, published by Rai Ansar.