On July 28, 2026, Elon Musk told Vercel CEO Guillermo Rauch on X that xAI will ship Grok 4.6 around August 7 with 1.5 trillion parameters, followed a few weeks later by a 2.1T Grok 4.7. For developers tracking frontier model release cadence and coding agent selection, this guide covers the Grok 4.5 to 4.6 to 4.7 timeline, spec and competitor comparison tables, SFT/RL post-training breakdown, four red flags, the August 2026 release bottleneck, a six-step pre-launch evaluation checklist, and 5 FAQs. Bottom line: everything here comes from a single X reply — no official model card or third-party benchmarks yet. Verify against xAI channels before you commit.
Just one month after Grok 4.5 shipped, xAI is already previewing two larger models — which means teams evaluating Cursor/Grok agent workflows face a very short selection window. The timeline below separates confirmed events from official previews.
| Date | Event |
|---|---|
| 2026-07-08 | xAI ships Grok 4.5, its coding and agentic flagship co-trained with Cursor on real developer sessions. 500K-token context, pricing $2/$6 per million tokens, published model card with 15 benchmark scores. |
| 2026-07-16 | Moonshot AI's Kimi K3 goes live as a hosted service |
| 2026-07-26 | Kimi K3 releases full open weights a day early — 2.8T parameters, 1M-token context |
| 2026-07-28 | Musk posts the Grok 4.6/4.7 roadmap in reply to @rauchg — primary source for this article |
| ~2026-08-07 | Grok 4.6 target release — 1.5T parameters, SFT/RL upgrade focus |
| ~late Aug–early Sep 2026 | Grok 4.7 planned follow-up — 2.1T parameters; Musk says better than 4.6 in every way except slightly slower serving, with better token efficiency |
Treating a tweet as an official announcement: Musk timelines have historically slipped — "Musk time" often means days to two weeks late
Parameter count is not capability: 4.6 emphasizes SFT/RL, not raw scale; 4.7 is bigger but slower — two SKUs, two trade-offs
Competitive window is razor-thin: Grok 4.6 was previewed roughly 10 days after Kimi K3's open-weight shock; August may also bring Claude Fable 5.1 rumors
Benchmark scores are blank: Unlike Grok 4.5's detailed model card, 4.6 has zero independent evaluations
Pricing is undisclosed: Grok 4.5's $2/$6 is a reference only — next-gen pricing may shift
Vendor risk context: xAI sued a user in July over CSAM deepfakes; enterprises should factor supplier safety posture into selection
Musk time reminder: Timelines from Musk have historically slipped by days to weeks. Treat "around August 7" as a target, not a guarantee. Confirm via xAI's official blog or @xai before treating this article as a spec commitment.
| Model | Date | Parameters | Focus | Status |
|---|---|---|---|---|
| Grok 4.3 Beta | Apr 17, 2026 | Undisclosed | Prior baseline | Shipped |
| Grok 4.5 | Jul 8, 2026 | Undisclosed (single SKU, not MoE) | Coding/agentic, co-trained with Cursor | Shipped, benchmarked |
| Grok 4.6 | ~Aug 7, 2026 | 1.5T | SFT/RL upgrade | Announced via tweet, unshipped |
| Grok 4.7 | ~late Aug–early Sep 2026 | 2.1T | Broad upgrade over 4.6, better token efficiency | Announced via tweet, unshipped |
Note: Parameter counts, pricing, and benchmark scores are vendor/Musk claims. Grok 4.6 and 4.7 have no third-party evaluations yet.
| Model | Vendor | Parameters | Context | Pricing (input/output per 1M tokens) | Source |
|---|---|---|---|---|---|
| Grok 4.5 | xAI | Undisclosed | 500K | $2 / $6 | xAI official |
| Grok 4.6 (announced) | xAI | 1.5T | Undisclosed | Undisclosed | Musk's X post (unverified) |
| Kimi K3 | Moonshot AI | 2.8T (MoE, ~16/896 experts active) | 1M | $0.30 (cache hit) / $3 input, $15 output | Moonshot + Hugging Face |
| Claude Fable 5.1 (rumored) | Anthropic | Undisclosed | Undisclosed | Rumored unchanged from Fable 5 ($10 / $50) | 36kr, WinCentral leaks — unconfirmed |
| GPT-5.6 Sol | OpenAI | Undisclosed | Undisclosed | Undisclosed | OpenAI official |
Grok 4.6 and Claude Fable 5.1 rows are pre-release claims, not verified benchmarks — useful for release timing and positioning, not head-to-head performance comparisons. For developers, token efficiency and real task cost matter more than parameter count alone.
Supervised fine-tuning (SFT) shapes model behavior using curated example outputs; reinforcement learning (RL) uses reward signals to teach which action sequences actually work — critical for multi-step agentic tasks. Musk emphasized "significantly improved SFT & RL" rather than raw scale, continuing Grok 4.5's playbook: real Cursor developer session data helped Grok 4.5 hit Terminal-Bench 2.1 83.3% and SWE-Bench Pro 64.7% while using roughly 15,954 output tokens per task versus Opus 4.8's 67,020 — a 4.2x efficiency gap.
Grok 4.6's jump to 1.5T is a real scale increase, but Musk's framing of Grok 4.7 — bigger at 2.1T, "better in every way except slightly slower to serve" — suggests xAI is deliberately building two SKUs with different latency-vs-quality trade-offs, similar to Anthropic's Sonnet/Opus split or OpenAI's mini/full tiers.
Grok 4.6's target lands roughly 10 days after Kimi K3's full open-weight release. Kimi K3 topped the Frontend Code Arena leaderboard at 1,679 points — the first open-weight model to beat every closed model on that board — and ranked third on Artificial Analysis's Intelligence Index. Musk himself called K3 "impressive" in benchmark comment threads. The most plausible read: xAI is compressing its cadence to three frontier models in roughly two months because competition from OpenAI, Anthropic, and Chinese labs intensified sharply in July.
Lock your information sources: Follow xAI's blog (x.ai/news) and @xai — don't base procurement on second-hand media alone
Establish a baseline: Record Grok 4.5 token usage and pass rates on your team's real tasks for A/B comparison after 4.6 ships
Buffer your budget: Estimate upper bounds using 4.5's $2/$6 pricing, and prepare fallback routes to Kimi K3 or GPT-5.6 Sol
Separate 4.6 from 4.7: Latency-sensitive workloads may favor 4.6; teams needing peak capability who can accept slower inference should watch 4.7
Review vendor risk: Factor xAI's July CSAM lawsuit and Common Sense Media's child-safety rating into enterprise policy review
Prepare access paths: Based on Grok 4.5, check Grok Build, xAI API, and Cursor for day-one integration announcements
Industry split context: Some labs call for a slower pace while xAI compresses its release cycle to three frontier models in about two months. For teams, token efficiency and real task cost are more durable decision criteria than parameter count or leaderboard rank alone.
If Musk's timeline holds, Grok 4.6 and Grok 4.7 will land in the same month as rumored Claude Fable 5.1 (per 36kr and WinCentral leaks, timed to beat OpenAI's anticipated GPT-6) and just weeks after Kimi K3's open-weight shock. Selection windows compress to weeks — last month's flagship may face next month's competitor before you've finished evaluation.
If you plan to wire Grok 4.6 into Cursor-style long-session coding agents or iOS CI pipelines, running CLI agents on a laptop or unstable Linux VPS often means memory pressure, dropped sessions, and missing Xcode/Metal toolchain support. For production workloads needing stable SSH sessions, DerivedData caching, and iOS CI/CD automation, NodeMini's Mac Mini cloud rental is usually the better fit — dedicated nodes, second-scale provisioning, agents and builds on the same real Mac hardware. See Mac Mini rental rates.
Information current as of July 30, 2026. Sources: xAI Grok 4.5 official announcement and model card, Musk's July 28, 2026 X post, Moonshot Kimi K3 official release, 36kr/WinCentral on Claude Fable 5.1 rumors, The Verge/TechTimes on "Pacing the Frontier," Ars Technica/TechCrunch on xAI safety controversies.
Musk said "around August 7, 2026" in an X post, but xAI has not officially confirmed a date. Treat it as a target that could shift. Once it ships, see Mac Mini rental rates for stable agent hosting.
Grok 4.6 is a 1.5T-parameter model focused on SFT/RL post-training improvements. Grok 4.7, expected a few weeks later, is a larger 2.1T model that Musk says outperforms 4.6 across the board except for serving speed, where it trades some latency for better token efficiency.
Too early to tell. Grok 4.6 has no published benchmarks yet, Kimi K3 already has verified third-party scores (including a leaderboard-topping result on Frontend Code Arena), and Claude Fable 5.1 hasn't even been officially confirmed by Anthropic. A real comparison isn't possible until Grok 4.6 ships with a model card.
Unknown. Grok 4.5 launched at $2 per million input tokens and $6 per million output tokens, which is a reasonable reference point, but xAI hasn't disclosed Grok 4.6 pricing. Always confirm on release day.
Based on Grok 4.5's rollout, expect Grok Build, the xAI API, and the xAI console to get access first, with third-party integrations (like Grok 4.5's day-one Cursor availability) following shortly after. For setup questions, see the help center.