Grok 4.6 & 4.7: Release Date, Size & What's Confirmed

Grok 4.5 launched publicly on 8 July 2026, built on a 1.5-trillion-parameter foundation model xAI calls V9. Just ten days later, on 18 July, Musk posted that the next model had already finished training. Six days after that, on 24 July, he attached hard dates: “Grok 4.6 in 2 weeks and Grok 4.7 in 4 weeks.” That compresses two full-generation iterations into a single calendar month, a cadence no other frontier lab has matched publicly in 2026.

• Confirmed: Musk's two X posts and their exact wording (18 and 24 July 2026).
• Confirmed: Grok 4.5's real spec, as the baseline: 1.5T V9 foundation, 500K context, $2/$6 per million tokens.
• Not confirmed: Grok 4.6/4.7 parameter counts, context window, pricing, benchmarks, or an official xAI release date.
• Not confirmed: Whether the “2 weeks / 4 weeks” timeline will hold; xAI has slipped founder-stated deadlines on prior releases.
• Stated goal (Musk, not a benchmark): match or exceed Kimi K3's capability while keeping Grok 4.5's speed and token efficiency.

The full public record on Grok 4.6 and 4.7, as of this article's publication, is two posts from Elon Musk's personal X account. The first, on 18 July 2026, claimed that training on xAI's next model had already completed, just ten days after Grok 4.5's public rollout. The second, on 24 July 2026, added the specific commitment that outlets have run with since: “Grok 4.6 in 2 weeks and Grok 4.7 in 4 weeks.”

Neither post came with a benchmark, a spec sheet, or a link to an xAI blog post or technical report. Both are exactly what they appear to be: founder social-media posts, not product announcements. xAI itself has not, at the time of writing, published a corresponding statement, roadmap page, or model card referencing either Grok 4.6 or Grok 4.7 by name.

It is worth being precise about the difference between “Musk said it” and “xAI confirmed it,” because most Grok 4.6/4.7 coverage online blends the two without a clear seam:

• Confirmed, by Musk directly: a training-complete claim (18 July) and a two-model, one-month release timeline (24 July).
• Confirmed, by xAI's actual Grok 4.5 launch: the 1.5T V9 foundation model, 500K token context, $2/$6 per million token pricing, and real, independently measured Artificial Analysis benchmark scores, the only hard numbers in this entire story.
• Not confirmed, by anyone official: Grok 4.6 or 4.7's parameter count, architecture, context window, pricing, training data composition, or benchmark performance.

Every parameter-count figure, every benchmark projection, and every specific release-date estimate you encounter beyond those two posts is third-party analysis layered on top, not additional disclosure from xAI. That includes this article's own estimates further down, which are explicitly flagged as such.

Because there is no Grok 4.6 benchmark to report, the only honest way to size up the claim is against what Grok 4.5 actually, verifiably does today. On the Artificial Analysis Intelligence Index, Grok 4.5 (high) scores 54, fourth place among tracked frontier models, behind Claude Fable 5, GPT-5.5 and Claude Opus 4.8. On the Coding Agent Index it scores 76, on par with GPT-5.5 in the Codex harness.

Grok 4.5's standout number was never raw intelligence, it was efficiency: $0.31 per Intelligence Index task, roughly a sixth of Opus 4.8's cost, achieved by using 1.9 million average tokens per coding-agent task against 6.2 million for GPT-5.5 and 7.2 million for Fable 5. That efficiency framing, not a raw capability crown, is exactly what Musk has now said Grok 4.6 needs to preserve while adding more capability on top, which is a materially harder engineering target than simply scaling parameters.

Grok 4.5 also carries a specific, documented weakness worth restating here because it directly bears on how much scrutiny 4.6 deserves at launch: on Artificial Analysis' AA-Omniscience Index, which tests whether a model knows what it doesn't know, Grok 4.5's accuracy improved from 35% to 52% versus Grok 4.3, but its hallucination rate rose even faster, from 25% to 54%. xAI has still not published a Grok 4.5-specific model card addressing that regression. Whether Grok 4.6 repeats, worsens or fixes that pattern is exactly the kind of question a proper launch document would need to answer, and exactly the kind of question a two-sentence X post cannot.

Several outlets covering Musk's posts have attached a parameter estimate to Grok 4.6: roughly 2 trillion parameters, up from Grok 4.5's confirmed 1.5-trillion V9 foundation, a 33% increase. Grok 4.7 is reported, again by third parties rather than xAI, to be larger still, somewhere in a 4-to-6-trillion-parameter range.

Neither figure has been stated by xAI or by Musk in the posts this story is based on. They are analyst and outlet estimates, extrapolated from xAI's prior scaling pattern (Grok 4.3's V8-series architecture to Grok 4.5's V9, roughly tripling scale) rather than disclosed specifications. Treat them exactly as that: a plausible guess based on precedent, not a leak.

The one specific competitive target Musk has actually named is Kimi K3, Moonshot AI's open-weight model that has drawn intense attention in July 2026, both for its capability and for the US-China export-control dispute surrounding it. The stated framing: Grok 4.6 should “match or exceed Kimi K3's capability while keeping the generation speed and token efficiency of the smaller Grok 4.5.”

Naming a specific open-weight rival as the bar to clear, rather than simply claiming a new leaderboard-topping score, is notable framing in itself: it positions Grok 4.6 as a response to a competitor that shipped its weights openly and drew real developer attention, rather than as a pure internal capability milestone. It also means Grok 4.6's eventual reception will likely be judged directly against Kimi K3's independently verified benchmark numbers, whatever those look like by the time Grok 4.6 actually ships.

There is a second layer to why Kimi K3 specifically makes for a pointed target: its release has been entangled in a US-China export-control dispute over the Nvidia hardware used to train it, covered in detail in our separate piece on the story. A closed-weight, US-based lab explicitly naming an openly-weighted Chinese model as the capability bar to clear is itself a small data point in the broader 2026 narrative around open-weight models closing the gap to closed frontier labs faster than expected, a theme that has come up repeatedly across this year's releases from Moonshot, Alibaba's Qwen team, Zhipu's GLM line and DeepSeek, alongside the closed-lab frontier represented by Anthropic, OpenAI, Google and xAI itself.

Grok 4.6 and 4.7 would extend a lineage that started with Grok 1, released in November 2023 as xAI's first public model, followed by Grok 1.5 in early 2024 and the multimodal Grok 1.5V shortly after. Grok 2 shipped in August 2024 alongside an image-generation feature built on Black Forest Labs' Flux, and Grok 3 launched in February 2025 as xAI's first model explicitly pitched as frontier-competitive, trained on the Colossus supercomputer cluster in Memphis. Grok 4 followed in July 2025, with Grok 4 Fast arriving that September as a cheaper, faster variant, Grok 4.1 in November 2025, and Grok 4.3 in the following months, before Grok 4.5's 8 July 2026 launch on the V9 foundation.

That history shows two things relevant to the Grok 4.6/4.7 claim. First, xAI has genuinely shortened its release intervals over time, from roughly annual major generations in 2023 to sub-monthly point releases by mid-2026, so a fast follow-up to 4.5 is consistent with the company's trajectory rather than a one-off anomaly. Second, nearly every one of those releases (Grok 3, Grok 4 and Grok 4.5 especially) launched with an accompanying model card or technical write-up, which makes Grok 4.5's launch without one, and the current total absence of any equivalent documentation for 4.6/4.7, a genuine departure worth watching rather than the new normal.

xAI has built its identity around release speed since Grok 4 first shipped in mid-2025: Grok 4 Fast followed that September, Grok 4.1 in November, Grok 4.3 in the following months, and Grok 4.5 on 8 July 2026. Compressing Grok 4.6 and 4.7 into the four weeks immediately after 4.5 would be the fastest back-to-back cycle in that entire history, and, per multiple outlets covering the announcement, faster than any other frontier lab has managed publicly in 2026.

That speed has a documented cost: xAI, and Musk personally, have a track record of announcing ambitious dates on X that slip in practice, a pattern independent coverage of this very story explicitly flags. Nothing about the 18 and 24 July posts changes that history. The claims are plausible given xAI's genuine shipping velocity over the past year, but “plausible given past velocity” and “confirmed date” are different things, and this story currently rests entirely on the former.

Taking Musk's 24 July post literally, Grok 4.6 would land in early-to-mid August 2026 and Grok 4.7 roughly two weeks later, both before the end of August. Independent outlets covering the announcement, citing xAI's history of slipping self-imposed deadlines, suggest a more realistic window that pushes into September 2026, particularly for Grok 4.7, which was already described as “weeks behind” 4.6 in the original framing before any slippage is factored in.

Neither estimate should be read as more than an informed guess layered on top of a founder's social media post. The single most reliable signal will be xAI's own model card or blog post at actual launch, not the interval between now and Musk's stated dates.

xAI's compressed timeline lands in the same few weeks as several other major moves: Claude Opus 5 shipped from Anthropic, Google confirmed Gemini 4 pre-training has begun (covered in our companion article), GPT-5.6 continues its rollout from OpenAI, and Kimi K3's open-weight release from Moonshot AI is the explicit benchmark xAI has named for Grok 4.6 to beat. Every major lab is, in effect, either shipping or pre-announcing its next move in the same narrow window, which is the real context for why xAI is racing to ship twice in one month rather than sitting on Grok 4.5's current, genuinely strong price-to-performance position.

That density of simultaneous announcements also means any Grok 4.6 benchmark, once it actually exists, will be judged against a moving target: whichever of Kimi K3, Opus 5, GPT-5.6 and (much further out) Gemini 4 has the strongest independently verified numbers at the moment Grok 4.6 ships, not against today's leaderboard.

It is worth sitting with how unusual this particular week is even by 2026's already-fast standards: five separate labs (xAI, Anthropic, Google, OpenAI and Moonshot) each had live, newsworthy model developments inside the same seven-to-ten-day span. For buyers and developers, that compresses the useful shelf life of any single “best model” claim to a matter of weeks, which is itself a reasonable part of why Musk is framing Grok 4.6's target as beating a named, currently-shipping rival (Kimi K3) rather than an abstract leaderboard position that will likely have shifted again by the time 4.6 actually ships.

The gap between a founder's tweet and a shipped model with a published spec is where most of the real uncertainty in this story lives. Specifically unresolved:

• Will xAI publish a model card this time? Grok 4.5 shipped without one, a gap flagged in independent coverage and a departure from the model cards Grok 4, 4 Fast and 4.1 all received.
• What will pricing look like? Grok 4.5's $2/$6 per million tokens undercut every other frontier lab; whether that holds for a larger, more capable 4.6/4.7 is unstated.
• Will the token-efficiency claim survive a real scale-up? Historically, larger models cost more per task even when they are more capable; Musk's framing implies xAI expects to defy that pattern.
• Does the AA-Omniscience hallucination-rate regression seen in Grok 4.5 (which more than doubled versus Grok 4.3, from 25% to 54%) continue, worsen, or get addressed in 4.6/4.7? No safety-evaluation commitment has been made either way.
• Will both dates hold, slip together, or slip independently? Grok 4.7 was already described as trailing 4.6 by weeks before any slippage; a delay to 4.6 would presumably cascade.

Teams currently evaluating Grok 4.5 for production agentic or coding workloads have no reason to pause on Grok 4.6's account: 4.5 is shipped, priced and independently benchmarked today, while 4.6 has none of those things yet. Anyone tracking the Kimi K3 story should note that xAI has now explicitly named it as a competitive target, which is likely to keep Kimi K3 benchmark comparisons relevant well beyond its own launch cycle. Compliance and procurement teams should specifically watch whether xAI publishes a model card at Grok 4.6's actual launch, given the gap left by Grok 4.5's release, rather than assuming a safety evaluation will simply appear.

Journalists and analysts citing this story should be careful to attribute the timeline correctly: it is accurate to say “Musk stated Grok 4.6 would ship in two weeks,” it is not accurate to say “xAI confirmed a release date,” and the distinction matters given how often the two get collapsed into one headline once a founder posts a number on X. Retail investors and SpaceXAI-adjacent readers should also note that a shipped-on-schedule Grok 4.6 and 4.7 would be a genuine, positive proof point for xAI's post-merger execution speed, which is presumably part of why Musk is stating the timeline publicly rather than letting it leak.

Grok 4.6 and Grok 4.7 are real, in that xAI is very likely training both right now, and Musk's own account of a finished training run on 18 July is a specific, checkable claim rather than vague hype. But “in training” and “shipping in two and four weeks” are two different levels of certainty, and this entire story currently rests on two X posts with no accompanying xAI documentation of any kind.

The right way to treat this story today is as a confirmed roadmap direction with an unconfirmed timeline: xAI is racing to answer Kimi K3 and keep pace with Opus 5, GPT-5.6 and Gemini 4's pre-training, and it plans to do it faster than any rival lab has managed. Whether “2 weeks and 4 weeks” turns out to mean August or September, and whether xAI finally ships a model card alongside it, are the two things actually worth watching for next, not the specific numbers currently circulating as fact. This article will be updated as soon as xAI publishes an official Grok 4.6 or 4.7 announcement.