What is Grok 4.3 and why does it matter?
Grok 4.3 is xAI's newest large language model — a beta update to the Grok 4-series, released silently on April 17, 2026 with no press post, no livestream, and no official xAI blog. The model surfaced in the model selector at grok.com and the iOS / Android Grok apps for users on the SuperGrok Heavy tier ($300/mo), where it remains gated as of April 24, 2026.
The headline change in grok-4.3 is native video input — the first Grok model to accept video uploads directly in chat, summarize specific timestamps, and reason about footage. Independent reviewers at TechSifted and BuildFastWithAI also confirmed Grok 4.3 can generate downloadable PDFs, populated spreadsheets, and PowerPoint decks directly inside chat — capabilities that did not exist in Grok 4.20.
Architecturally, Grok 4.3 carries forward the 2,000,000-token context window and 16-agent Heavy multi-agent reasoning system from Grok 4.20. Early testers describe the change as “more compute per agent, not more agents,” per theaipixel. xAI has not published any benchmark scores for 4.3, so this page deliberately quotes none — only tester impressions and reviewer commentary.
What ships in Grok 4.3 Beta?
Four documented additions, every one cited inline. No unsourced benchmark numbers — xAI has published none for 4.3.
Grok 4.3 Beta accepts video uploads directly in chat — ask questions about a clip, summarize a specific timestamp, or extract information from footage. The first Grok model with native video understanding.
ChatlyAI: Grok 4.3 Beta video, slides & speech ↗Generates downloadable PDFs, populated spreadsheets, and PowerPoint decks directly inside the chat surface — capabilities not present in Grok 4.20.
BuildFastWithAI: Grok 4.3 review ↗Grok 4.3 carries forward the 2M-token context window introduced in Grok 4.1 Fast and Grok 4.20 — still the largest among Western closed-weight frontier models as of April 2026.
TechSifted: Grok 4.3 review (Apr 2026) ↗The 16-agent Heavy multi-agent reasoning system from Grok 4.20 carries into 4.3, with reportedly more compute per agent rather than more agents — early testers note deeper reasoning.
BuildFastWithAI: 16-agent reasoning ↗Grok 4.3 vs Grok 4.20 vs Grok 4
The honest line-by-line — what 4.3 actually changes, and what it carries over unchanged.
| Feature | Grok 4.3 Beta | Grok 4.20 | Grok 4 |
|---|---|---|---|
| Released | Apr 17, 2026 | Mar 2026 | Jul 9, 2025 |
| Native video input | Yes | No | No |
| PDF / slide / sheet output | Yes | No | No |
| Context window | 2,000,000 | 2,000,000 | 256,000 |
| Heavy (multi-agent) | 16-agent | 16-agent | 16-agent |
| Available to | SuperGrok Heavy ($300/mo) | SuperGrok ($30/mo)+ | X Premium+ ($22/mo)+ |
| Status | Beta — silent rollout | GA | GA |
| xAI release post | None | Yes | Yes |
The practical rule, given the silent rollout: Grok 4.3 is the most interesting Grok if you specifically need video input and you're already on SuperGrok Heavy. For everyone else, Grok 4 is the stable frontier choice today, with verified Humanity's Last Exam (25.4%), ARC-AGI-2 (16.2%), and GPQA Diamond (87.5%) scores published by xAI. See deeper breakdowns at the full Grok family, Grok 3, and the broader ChatGPT-alternative landscape.
What did early testers say about Grok 4.3?
No xAI release blog exists. The signal is reviewer commentary, all attributed and linked.
“xAI dropped Grok 4.3 Beta with no press release, no announcement post — it just appeared silently in the model selector on grok.com on April 17, 2026. The headline addition is native video input.”
“Grok 4.3 can now generate downloadable PDFs, fully populated spreadsheets, and PowerPoint decks directly from conversation — capabilities that weren't there in 4.20.”
“If the multi-agent architecture from 4.20 (16 agents in Heavy mode) is the engine, 4.3 appears to have more compute per agent, not just more agents — early testers note improved reasoning depth.”
Flash news coverage from Phemex and KuCoin corroborated the silent rollout on April 17, 2026, and Releasebot's xAI feed logs the version bump without an accompanying changelog.
How can you actually use Grok 4.3 today?
Honest answer: you almost certainly can't — yet. Grok 4.3 Beta is gated to SuperGrok Heavy at $300/month per Grokipedia's SuperGrok Heavy entry. Standard SuperGrok subscribers ($30/mo) can see the model name in the dropdown but the request returns an access-denied error. xAI has not announced an API release date.
The fastest no-waitlist path to a frontier Grok experience is ZeroTwo: open multi-model chat, pick Grok 4 from the model selector, and prompt. ZeroTwo Pro at $29.99/month bundles Grok 4 with GPT-5, Claude Opus 4.6, and Gemini 2.5 Pro — less than X Premium+, with no SuperGrok Heavy required.
When Grok 4.3 reaches the xAI public API — most reviewers expect mid-to-late May 2026 — ZeroTwo will add it the same day, and existing Pro subscribers will get it included. Until then, see our deeper guides on running Grok free on ZeroTwo and the GPT-5 deep-dive.
What to remember about Grok 4.3
- Grok 4.3 Beta is xAI's newest model — released silently April 17, 2026, gated to SuperGrok Heavy ($300/month).
- Headline change vs Grok 4.20: native video input — upload a clip and reason about specific timestamps conversationally.
- Also new in 4.3: downloadable PDF, PowerPoint slide, and spreadsheet generation directly inside chat.
- The 2,000,000-token context window and 16-agent Heavy reasoning system carry over from Grok 4.20 unchanged.
- No official xAI release post and no public benchmark scores as of Apr 24, 2026 — beta numbers are tester impressions only.
- Run Grok 4 today on ZeroTwo at $29.99/month — no SuperGrok Heavy, no waitlist, side-by-side with GPT-5 and Claude.
Frequently asked about Grok 4.3
Nine direct answers, sourced from independent reviewers — xAI has not published an official Grok 4.3 release post. Run Grok 4 on ZeroTwo today while you wait for 4.3 to leave SuperGrok Heavy.
What is Grok 4.3?
Grok 4.3 Beta is xAI's newest large language model, released April 17, 2026. The headline change vs Grok 4.20 is native video input — you can upload a clip and reason about it conversationally — plus the ability to generate downloadable PDFs, PowerPoint decks, and populated spreadsheets directly inside chat. Grok 4.3 keeps the 2,000,000-token context window and 16-agent Heavy reasoning system from Grok 4.20. It is currently in beta and gated to the SuperGrok Heavy plan at $300 per month, with broader rollout expected in mid-to-late May 2026 per multiple independent reviewers.
When was Grok 4.3 released?
Grok 4.3 Beta was released on April 17, 2026. There was no official xAI blog post, livestream, or press release — the model appeared silently in the model selector on grok.com and the iOS / Android Grok apps. Independent coverage on the same day from TechSifted, BuildFastWithAI, ChatlyAI and theaipixel substack confirmed the rollout to SuperGrok Heavy subscribers.
How do I access Grok 4.3?
As of April 24, 2026, Grok 4.3 Beta is gated to the SuperGrok Heavy subscription at $300 per month — the same tier required for Grok 4 Heavy. Standard SuperGrok subscribers ($30/mo) can see the model name in the dropdown but cannot access it yet. xAI has not published an API release date. The fastest no-waitlist path to a frontier Grok experience is ZeroTwo, which runs Grok 4 today alongside GPT-5, Claude Opus, and Gemini at $29.99/month — no SuperGrok Heavy required.
What's new in Grok 4.3 vs Grok 4.20?
Three concrete additions per independent reviewers: (1) native video input — upload a clip and ask questions about specific timestamps; (2) downloadable PDF, slide, and spreadsheet generation directly in chat; (3) deeper per-agent reasoning in the 16-agent Heavy system, though the agent count is unchanged. The 2,000,000-token context window and Heavy architecture carry over from 4.20 unchanged.
Are there official Grok 4.3 benchmarks?
No. As of April 24, 2026, xAI has not published any benchmark scores for Grok 4.3 Beta. There is no release blog, no model card, and no Artificial Analysis or LMArena entry yet. Numbers floating in independent reviews are tester impressions, not validated evaluations. This page deliberately does not quote unsourced 4.3 benchmark numbers — for verified Grok benchmarks, see the Grok 4 page where xAI did publish HLE 25.4%, ARC-AGI-2 16.2%, and GPQA 87.5% scores.
How does Grok 4.3 compare to GPT-5.5 and Claude Opus 4.6?
On paper, Grok 4.3's video-native input and 2M context window are competitive with GPT-5.5 and Claude Opus 4.6. In practice, no apples-to-apples benchmark comparison exists yet because xAI has not released scores for 4.3, and the model is restricted to SuperGrok Heavy subscribers — limiting independent evaluation. The honest answer: wait for Artificial Analysis and LMArena to score 4.3 in May 2026, or run all three side-by-side on ZeroTwo when 4.3 reaches the API.
What is SuperGrok Heavy?
SuperGrok Heavy is xAI's top consumer subscription tier, launched July 9, 2025 at $300 per month. It originally unlocked Grok 4 Heavy's 16-agent multi-agent mode, the highest rate limits, and priority access. As of April 17, 2026, SuperGrok Heavy is also the only consumer surface where Grok 4.3 Beta is available — and the only way to use Grok 4.3 today.
Can I try Grok without SuperGrok Heavy?
Yes. ZeroTwo includes Grok 4 — xAI's frontier GA model — on the free tier and unlimited on Pro at $29.99/month, alongside GPT-5, Claude Opus 4.6, Gemini 2.5 Pro, and 60+ other frontier models. You can run Grok 4 side-by-side against any other frontier model on the same prompt. While Grok 4.3 specifically is still locked behind $300/mo SuperGrok Heavy, ZeroTwo is the fastest no-waitlist way to get a frontier Grok experience right now.
Is Grok 4.3 stable enough for production?
No. Grok 4.3 is in beta — xAI's silent rollout pattern, the lack of a release blog, the lack of public benchmark numbers, and the SuperGrok-Heavy-only gating all indicate a controlled-rollout stage. Multiple independent reviewers expect a full rollout to standard SuperGrok in mid-to-late May 2026. For production workloads today, Grok 4 GA, GPT-5, and Claude Opus 4.6 are the stable frontier choices — all three are available unlimited on ZeroTwo Pro.
The ZeroTwo editorial team tracks every frontier model release, runs benchmark comparisons across 60+ models, and updates these pages with primary-source numbers — never marketing copy.
Published · Updated
Skip the $300 waitlist.
Run Grok 4, GPT-5, Claude Opus 4.6, and Gemini 2.5 Pro side-by-side on ZeroTwo — same prompt, four frontier models, one subscription. No SuperGrok Heavy required.
Try Grok 4 on ZeroTwo today