Grok 4.3 Beta: Access, Pricing, Rate Limits and SuperGrok Rollout

xAI expanded Grok 4.3 beta access on April 19, 2026 from SuperGrok Heavy ($300/mo) only to include X Premium+ ($40/mo) and SuperGrok ($30/mo) subscribers. The broader rollout lands 48 hours after the initial April 17 Heavy-gated launch.

Fast answer

Grok 4.3 beta is no longer limited to SuperGrok Heavy. As of the April 19 rollout, X Premium+, SuperGrok, and SuperGrok Heavy users can access the beta. Heavy remains the high-limit, priority tier; Premium+ and SuperGrok are the cheaper entry points for trying the 4.3 model.

Who gets it now

Rate limits: what is known

Heavy still gets priority access, higher rate limits, and 16-agent architecture. Premium+ and SuperGrok get access to the 1T-parameter model but with standard beta limits.

xAI has not published exact per-tier Grok 4.3 beta message caps in the sources reviewed for this article. Treat any specific cap screenshots as volatile until xAI publishes a stable tier table.

Rollout pattern matches Grok 4.20 exactly

xAI’s beta playbook:

1. Day 0: launch to the highest-tier Heavy audience. Signals exclusivity. Drives some users to upgrade.
2. Day 2-3: expand to mid-tier SuperGrok + X Premium+. Broader feedback loop.
3. Day 7-14: typically removes the beta tag once stability holds, model ships to all paid tiers.
4. Day 30+: free tier access (possibly rate-limited) if the broader rollout is clean.

This is the exact pattern Grok 4.20 followed in Q1 2026. Today’s expansion is Day 2 of that cycle.

What Grok 4.3 brings

Six capabilities beyond Grok 4.20:

1. Native PDF generation - document creation directly in chat
2. PowerPoint slide output - deck generation from prompts
3. Spreadsheet output - CSV and XLSX structured data
4. Video input - multimodal vision with moving images
5. Sharper reasoning - benchmark improvements from extended training
6. Grok Computer integration - alongside the ongoing Computer Use rollout

Model scale: 1 trillion parameters, double Grok 4.20’s ~500B. On SuperGrok Heavy, a 16-agent Heavy architecture orchestrates multiple Grok 4.3 instances in parallel for complex tasks.

What Premium+ / SuperGrok users get vs Heavy

Premium+ ($40/mo) and SuperGrok ($30/mo) access includes:

• Grok 4.3 beta model access
• All six new output capabilities (PDF, PPT, XLSX, video input, sharper reasoning, Grok Computer)
• 2M token context window

SuperGrok Heavy ($300/mo) adds on top:

• 16-agent Heavy architecture (parallel multi-agent orchestration)
• Higher rate limits
• Priority queue access during load spikes
• Early access to any post-4.3 previews

If you’re evaluating Grok 4.3 specifically for output capabilities (decks, docs, video), Premium+ or SuperGrok is now the right starting point. Heavy only makes sense if you need the multi-agent workload or predictable rate limits at volume.

Where this sits in the frontier model landscape

April 2026 frontier comparison:

Grok 4.3 is the only frontier model with a disclosed 1T-parameter headline figure, though comparable effective compute to GPT-5.4 and Opus 4.7 is unverified. The 2M context window is currently the largest on a major commercial chatbot.

What to watch

• Beta tag removal date. If the pattern holds, Grok 4.3 drops beta and ships to all paid tiers within 7-14 days (late April to early May).
• Free tier rollout. Historical pattern: rate-limited free access 4-8 weeks after GA.
• Grok Imagine + Computer integration. Grok 4.3 lands with both parallel tracks. Expect tighter integration as the beta matures.
• Benchmark publications. xAI has been slower than Anthropic / OpenAI to publish independent benchmarks for Grok 4.3. Independent evals (ARC-AGI, SWE-bench, Artificial Analysis) expected within 2-3 weeks.

What this means

For Grok users: If you were considering upgrading to SuperGrok Heavy just to get 4.3 access, you can now get it on SuperGrok ($30/mo) or Premium+ ($40/mo) at 10% the cost. Heavy is now justified only by multi-agent architecture + priority rate limits.

For the frontier race: Grok 4.3 at $30/mo puts pressure on ChatGPT Plus and Claude Pro at the same price point. The 2M context and 1T-param positioning is a commercial differentiator regardless of raw benchmark gap.

For xAI: The 2-day rollout cycle (Heavy → Premium+ / SuperGrok) shows confidence in the model. Slower rollouts typically signal stability concerns; xAI is running the opposite playbook.

See the full Grok tool page for the current tier matrix and feature breakdown.