# Grok 4.6

> Source: https://aiwiki.ai/wiki/grok_4_6
> Summary: Grok 4.6 is a proprietary large language model and reasoning model in the Grok family, developed by SpaceXAI and released jointly with Cursor through the xAI API on August 12, 2026.
> Updated: 2026-09-23
> Fact-checked: 2026-09-23
> Categories: AI Models, Generative AI, Large Language Models, Reasoning Models
> License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) - attribute to "AI Wiki (aiwiki.ai)"
> Cite as: AI Wiki. "Grok 4.6." aiwiki.ai, 23 Sept 2026. https://aiwiki.ai/wiki/grok_4_6
> From AI Wiki (https://aiwiki.ai), the free encyclopedia of artificial intelligence. Reuse freely with attribution.

| Field | Value |
|---|---|
| Developer | [SpaceXAI](https://aiwiki.ai/wiki/xai) (released jointly with [Cursor](https://aiwiki.ai/wiki/cursor)) |
| Release date | August 12, 2026 |
| Model identifier | `grok-4.6` |
| Predecessor | [Grok 4.5](https://aiwiki.ai/wiki/grok_4_5) |
| Successor | [Grok 4.7](https://aiwiki.ai/wiki/grok_4_7) (September 21, 2026) |
| Architecture | Not formally published; 1.5 trillion parameters per Elon Musk |
| Input and output | Text and image input; text output |
| Context window | 500,000 tokens |
| Knowledge cutoff | February 1, 2026 |
| Reasoning control | Low, medium, high (default), or xhigh; cannot be disabled |
| Pricing | $2 per million input tokens, $6 per million output tokens (doubled at 200,000+ input tokens) |
| Availability at launch | xAI API, [Grok Build](https://aiwiki.ai/wiki/grok_build), Cursor, model gateways |
| License | Proprietary |

**Grok 4.6** is a proprietary [large language model](https://aiwiki.ai/wiki/large_language_model) and [reasoning model](https://aiwiki.ai/wiki/reasoning_models) in the [Grok](https://aiwiki.ai/wiki/grok) family, developed by [SpaceXAI](https://aiwiki.ai/wiki/xai) and released jointly with [Cursor](https://aiwiki.ai/wiki/cursor) through the xAI API on August 12, 2026. It succeeds [Grok 4.5](https://aiwiki.ai/wiki/grok_4_5) at the same $2 and $6 per-million-token prices, keeps that model's 500,000-token [context window](https://aiwiki.ai/wiki/context_window) and February 2026 knowledge cutoff, and is positioned by both companies for coding, agentic tasks, and knowledge work, with what the launch post calls "a particular focus on long-running agents and more ambitious interactive and visual work."[1][3][4][8]

The release arrived five weeks after Grok 4.5 and closed a publicly telegraphed schedule: [Elon Musk](https://aiwiki.ai/wiki/elon_musk) had posted the model's timing, claimed parameter count, and training approach on X in late July, then repeated the schedule on SpaceX's first earnings call.[10][11][12] On [Artificial Analysis](https://aiwiki.ai/wiki/artificial_analysis)' Intelligence Index, the most-cited independent measure, Grok 4.6 at high reasoning effort scored 61 on the v4.1.1 index on launch day, up five points on Grok 4.5 and tied with OpenAI's GPT-5.6 Sol at maximum effort, behind only Anthropic's [Claude Opus 5](https://aiwiki.ai/wiki/claude_opus_5) and [Claude Fable 5](https://aiwiki.ai/wiki/claude_fable_5) configurations.[14][15] It was succeeded by [Grok 4.7](https://aiwiki.ai/wiki/grok_4_7) on September 21, 2026, which kept the same token prices, and it remained available in the xAI API afterward.[35][36][39]

## Announcement and release

The launch was unusual in how much of it Musk narrated in advance. On July 24, 2026, he posted a chart claiming that "Grok 4.5 and Opus 5 are alone on Pareto frontier," and replied to his own post: "Grok 4.6 in 2 weeks and Grok 4.7 in 4 weeks."[10] Four days later he added specifics: "Grok 4.6 releases around August 7. This will be the 1.5T model with significantly improved SFT & RL. Grok 4.7 will be the 2.1T model released a few weeks later."[11] On SpaceX's first earnings call as a public company, held August 4, 2026, he updated the schedule again, telling investors that Grok 4.6 would arrive the following week, that Grok 4.7 would follow three to four weeks after the call, and that Grok 5, expected around the end of the year, would incorporate SpaceX's own engineering data.[12][13]

The "around August 7" estimate passed without a release. Grok 4.6 shipped on August 12, 2026, announced at 15:32 UTC in a post from the @SpaceXAI account: "Introducing Grok 4.6. It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price."[9] SpaceXAI's release notes record API availability the same day, and Cursor published a matching launch post, opening "Today we are releasing Grok 4.6 together with SpaceXAI."[3][8] As with Grok 4.5, the model reached developer surfaces first: the announcement lists the xAI API, Grok Build (where it became the default model), Cursor on all plans, and the OpenRouter, [Vercel](https://aiwiki.ai/wiki/vercel), and [Cloudflare](https://aiwiki.ai/wiki/cloudflare) gateways.[1][4] Both companies offered double the included usage inside Grok Build and Cursor for the first week.[1][8]

The joint release continues the arrangement behind Grok 4.5, which Cursor and SpaceXAI trained together after SpaceX agreed to acquire Cursor's parent company [Anysphere](https://aiwiki.ai/wiki/anysphere); the corporate background is covered at [Grok 4.5](https://aiwiki.ai/wiki/grok_4_5) and [xAI](https://aiwiki.ai/wiki/xai). Cursor's Grok 4.6 post is published under the Cursor Team byline and its site still carried an "Anysphere, Inc." copyright notice on launch day.[8]

## Model and training

SpaceXAI has not formally published an architecture description or parameter count. The 1.5 trillion figure comes from Musk's July 28 post, which described Grok 4.6 as "the 1.5T model," the same size he had previously attached to Grok 4.5; his framing, and the launch post's, is that the gains over Grok 4.5 come from training rather than scale.[11][1] No model card accompanied the launch-day materials; for Grok 4.5, the model card followed six days after release.[1][8]

The launch post describes three stages of work on top of Grok 4.5's recipe. First, a "longer supplemental training run" used curated model-generated data for reasoning and advanced technical concepts, high-quality engineering data, and what the company calls an improved optimizer and training recipe. Second, SpaceXAI used Grok 4.5 itself to regenerate the [supervised fine-tuning](https://aiwiki.ai/wiki/sft) trajectories across reasoning efforts, agent harnesses, and domains such as STEM, software engineering, and knowledge work, filtering out problematic traces with model-based checks. Third, the model went through [reinforcement learning](https://aiwiki.ai/wiki/reinforcement_learning) on agentic tasks, including domain-specific environments for kernel optimization, web development, and computer-aided design.[1][8]

The documented interface matches Grok 4.5 in most respects: text and JPEG or PNG image input, text output, a 500,000-token context window, and a February 1, 2026 knowledge cutoff, with no access to real-time information unless a search tool is enabled.[4][5] Two things changed. The API imposes no text output limit, and the reasoning-effort control gained a fourth setting, xhigh, above the default high; SpaceXAI's documentation states that xhigh is available on grok-4.6, that requests sending it to grok-4.5 are treated as high, and that reasoning cannot be disabled.[3][4][7] The model works with the Responses and Chat Completions APIs and supports function calling plus SpaceXAI's server-side web search, X search, and code-execution tools.[4]

## Benchmarks

SpaceXAI's launch post says Grok 4.6 "achieves frontier intelligence across several agentic coding and knowledge work benchmarks" and that it matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index.[1] The announcement's own comparison table, reproduced below, is more mixed than that summary: Grok 4.6 leads on three of the ten rows, [Claude Fable 5](https://aiwiki.ai/wiki/claude_fable_5) on five, and GPT-5.6 Sol on two, with the Intelligence Index row going to Fable 5. The table compares Grok 4.6 at high reasoning effort against competitors at their maximum settings, and its footnote states that third-party scores are "the best of self-reported or publicly available results," so these are assembled figures rather than head-to-head runs.[1][2]

| Evaluation | Grok 4.6 (high) | Grok 4.5 (high) | GPT-5.6 Sol (max) | Claude Fable 5 (max) |
|---|---:|---:|---:|---:|
| AA Intelligence Index | 61 | 56 | 61 | **62** |
| [GDPval-AA](https://aiwiki.ai/wiki/gdpval) v2 | **1753** | 1526 | 1728 | 1741 |
| CursorBench v3.2 | 69.9% | 66.7% | 67.2% | **70.5%** |
| DeepSWE v1.1 | 65.9% | 54% | **73%** | 70% |
| FrontierCode v1.1 (Extended) | 61.3% | 56.6% | 60.6% | **63.6%** |
| APEX-Agents | 57.5% | 47.1% | 56.7% | **59.2%** |
| [Terminal-Bench](https://aiwiki.ai/wiki/terminal_bench) v3.0 | 26% | 15.7% | **34.6%** | 34.1% |
| APEX-SWE | 56.4% | 53.6% | not listed | **58.8%** |
| AA-Briefcase | **1577** | 1313 | 1502 | 1574 |
| Harvey LAB (Vals) | **15.8%** | 12.9% | 2.5% | 11.3% |

Source: SpaceXAI launch page as archived on August 12, 2026.[1][2] The benchmark graphic attached to the launch tweet differs from the web table in one cell, giving Claude Fable 5 64.9% rather than 63.6% on FrontierCode v1.1 (Extended); both appeared on launch day and the discrepancy was unexplained as of that evening.[9][2]

The gains over Grok 4.5 are largest on the knowledge-work rows: 227 points on GDPval-AA v2, 264 on AA-Briefcase, and a 10-point gain on Terminal-Bench v3.0, though that last benchmark remains the model's weakest result against competitors. The Harvey LAB row refers to Harvey's Legal Agent Benchmark, an evaluation of agents on legal work released by Harvey, whose live leaderboard is hosted by Vals AI and is scored on a held-out test set; it is the table's biggest lead and also its least symmetrical comparison, with GPT-5.6 Sol listed at 2.5%. Vals AI's hosted leaderboard lists Grok 4.6 at 15.83 percent, matching the vendor's figure, although on Vals' full board it ranks fifth, behind four Muse Spark configurations.[1][19]

### Independent measurement

Artificial Analysis evaluated Grok 4.6 at high reasoning effort on launch day and scored it 61 on Intelligence Index v4.1.1, a composite of nine evaluations including GDPval-AA v2, Terminal-Bench v2.1, [Humanity's Last Exam](https://aiwiki.ai/wiki/humanity_s_last_exam), and [GPQA](https://aiwiki.ai/wiki/gpqa) Diamond. That placed it sixth of 183 tracked models, tied with GPT-5.6 Sol at maximum effort and Claude Opus 5 at high effort, and behind Claude Opus 5 at maximum and xhigh effort (63) and Claude Fable 5 (62). Grok 4.5 (high) sat at 56 on the same board.[14][15]

The cost profile shifted against the predecessor even though the token prices did not. Artificial Analysis measured $0.84 per Intelligence Index task for Grok 4.6 against $0.36 for Grok 4.5, with the model generating 72 million output tokens across the index, around the median for comparable models; the evaluator's full run cost $1,068.[14][15] Grok 4.5's launch story had leaned heavily on below-median token use, so on this evidence the intelligence gain was partly bought with verbosity. Artificial Analysis had not yet published an output-speed measurement for Grok 4.6 as of August 12.[14]

Arena AI (formerly [LMArena](https://aiwiki.ai/wiki/lmarena)) listed grok-4.6 in four reasoning configurations (low, medium, high and xhigh) for head-to-head voting on launch day but had not yet ranked it in any modality.[16]

## Pricing and availability

Grok 4.6 costs $2 per million input tokens, $0.50 per million cached input tokens, and $6 per million output tokens for requests under 200,000 input tokens; at or above that threshold the rates double to $4, $1, and $12.[6] The headline input and output prices are identical to Grok 4.5's, supporting the launch tweet's "same price" claim, but the cached-input rate is higher: SpaceXAI's pricing page lists Grok 4.5 cached input at $0.30 ($0.60 long-context), so [prompt-cache](https://aiwiki.ai/wiki/prompt_caching)-heavy workloads pay more on the new model.[6][9] Both launch posts also mention "a fast variant" at "twice the price," without further specification.[1][8]

Server-side tools are billed separately, at $5 per 1,000 calls for web search, X search, and code execution, $2.50 per 1,000 for collections search, and $10 per 1,000 for file-attachment search.[6] SpaceXAI said at launch that from September 21, 2026 X search would move to $5 per 1,000 posts fetched and $10 per 1,000 user profiles fetched, replacing the per-call rate; the pricing page listed those per-post and per-profile rates when checked on September 23, 2026.[6][39] SpaceXAI recommends setting a `prompt_cache_key` so that a conversation's requests route to the same server, warning that cache hits are otherwise unreliable.[4]

At launch the model was available through the xAI API under the identifier `grok-4.6`, as the default model of the Grok Build coding agent, in Cursor on all plans, and through the OpenRouter, Vercel, and Cloudflare gateways; OpenRouter's listing records a 500,000-token context and a dated internal snapshot name of grok-4.6-20260810.[4][17] The launch materials do not mention availability in the consumer Grok apps or on X, the same staging Grok 4.5 followed.[1][3]

## Reception

Launch-day coverage was thin compared with Grok 4.5's, which had arrived alongside a $60 billion acquisition story. Reuters carried a brief wire item, "SpaceXAI Introduces Grok 4.6," and early trade coverage led with the Artificial Analysis tie with GPT-5.6 Sol.[18][20] The most substantive independent datapoint available on day one was Artificial Analysis' scoring described above, which supported the vendor's frontier claim on aggregate intelligence while showing the model's per-task cost more than doubling against its predecessor.[14][15]

The comparison targets Musk chose are worth noting: his July 24 chart measured Grok 4.5 against Claude Opus 5, and the launch table benchmarks Grok 4.6 against Claude Fable 5 and GPT-5.6 Sol at their maximum effort settings while Grok runs at high, one setting below its own xhigh.[10][1][7] With three frontier releases planned in roughly six weeks (4.5 in July, 4.6 in August, 4.7 promised weeks later), the schedule itself became part of the product story, and it did not hold exactly: the release landed five days after Musk's August 7 estimate.[11][12][3]

## Roadmap: Grok 4.7, Grok 4.8, and Grok 5

Musk's July 28 post described Grok 4.7 as "the 2.1T model released a few weeks later," saying it "will be better than 4.6 in every way, except slightly slower to serve"; on the August 4 earnings call he placed it three to four weeks out.[11][12] For Grok 5, the call guidance was a release around the end of 2026, trained with SpaceX's internal engineering and flight data, which Musk suggested could make it the strongest available model for engineering work.[12][13] At the time these were founder statements about unreleased products rather than documented specifications, and the Grok 4.6 schedule itself slipped five days between Musk's estimate and the release.[11][3] Grok 4.7 shipped on September 21, 2026; its release is summarized below.[35]

### The Grok 4.7 schedule

Grok 4.7 slipped repeatedly after Grok 4.6 shipped. The sequence below is taken from Musk's own posts; dates are UTC.

| Date | What Musk posted |
|---|---|
| July 21, 2026 | SpaceX's "massive corpus of world-class engineering data (excluding material blocked by ITAR) will be added during supplemental training of the 2T run," which he said would "dramatically improve Grok's engineering capabilities."[21] |
| July 28, 2026 | "Grok 4.7 will be the 2.1T model released a few weeks later."[11] |
| August 12, 2026 | "Grok 4.7 will exceed all current models," adding that "the SpaceX training corpus is so awesome & unique that I would be shocked if any model is better at real-world engineering than 4.7."[22] |
| September 2, 2026 | "Grok 4.7 comes out in 10 days."[23] |
| September 11, 2026 | "Grok 4.7 needs a few more days to cook," blamed on reinforcement learning that "might have penalized response length too much (or something)," leaving the model prone to giving up on hard tasks and not "sufficiently rigorous in checking its work."[24] |
| September 14, 2026 | Grok 4.7 "should be roughly on par with Opus 5.0, not 5.1. Better in some ways, worse in others. We need to fix multimodal performance."[25] |

The September 14 grade is a step down from the August 12 claim that the model would exceed everything then available; Musk's stated comparison point was Claude Opus 5.0, which he explicitly distinguished from 5.1.[22][25] The August 12 post was later deleted; X showed it as deleted by its author when checked on September 23, 2026.[22]

### Grok 4.8 and the C/C++ training stack

Musk named Grok 4.8 on September 14, 2026, at 01:24 UTC, in a reply to a user who had asked what to expect from SpaceXAI in September apart from Grok 4.7: "Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL."[26][27] Later the same morning he sketched the rest of the line: "Grok 4.8 will be a noticeable improvement. Grok 4.9 is probably Astra/Fable class. Grok 5 maybe better than anything. We shall see," a reference to OpenAI's [GPT-6 Astra](https://aiwiki.ai/wiki/gpt_6_astra) and Anthropic's Fable models.[25] Asked in a quote-post of that announcement how close Grok 4.8 would come to [artificial general intelligence](https://aiwiki.ai/wiki/artificial_general_intelligence), he answered: "That will be Grok 5."[28][29]

The software stack Musk attached to Grok 4.8 has its own trail of posts. On June 29, 2026, replying to a comment about aerospace engineers being put on Grok, he wrote that "truly massive gains will come in ~3 months when the entire training and inference stack is written in C/C++ and massively simplified (most software layers will be deleted completely) and we exact-map Grok to work incredibly well on a GB300," referring to [NVIDIA](https://aiwiki.ai/wiki/nvidia) hardware.[30] On September 14 he said the pre-training half of that plan had landed and the reinforcement-learning half had not: "Our training (aka pre-training) software is now the internally developed C/C++ stack, which was written primarily by the Starlink software team. This is awesome and probably best in industry. However, our reinforcement learning stack is still somewhat of a legacy pile of many different third-party software systems. This will get fixed, but isn't yet."[31] Replying later the same day to a commenter who guessed the stack was mostly AI-generated, he wrote: "Our C/C++ training stack was written by humans. AI is not yet good enough for extremely high performance software (but it will be)."[32]

Across the line, the parameter counts Musk has given rise from 1.5 trillion for Grok 4.6 to 2.1 trillion for Grok 4.7 and 2.5 trillion for Grok 4.8.[11][27] SpaceXAI's own Grok 4.7 launch post and documentation give no parameter count, saying only that the model uses "a new, larger base model compared to Grok 4.6."[35][36]

### Grok 4.7 release

As of September 15, 2026, neither Grok 4.7 nor Grok 4.8 had shipped. SpaceXAI's API model list still showed grok-4.6 as the newest Grok text model, with no grok-4.7 or grok-4.8 identifier, and the release notes carried no entry for either; the most recent entry was a September 2 notice about the retirement of an image model.[33][34]

Grok 4.7 arrived on September 21, 2026, nine days after the date implied by Musk's September 2 post.[23][36] SpaceXAI's release notes list it on the xAI API as `grok-4.7`, with the same 500,000-token context window, reasoning-effort settings and token rates as Grok 4.6: $2, $0.50 and $6 per million input, cached input and output tokens below 200,000 prompt tokens, and $4, $1 and $12 above.[36] The launch post says Grok 4.7 "uses a new, larger base model compared to Grok 4.6" and was trained with a longer reinforcement-learning run "weighted toward problems that take many hours to complete."[35] In the launch table, which sets Grok 4.7 at xhigh effort against Grok 4.6 at high, SpaceXAI reports 46.3% against 40.4% on CursorBench 4.0, 37.6% against 20.3% on Terminal-Bench 4.0 (the page originally said 38.0%, as did the launch image on X), and 1,657 against 1,546 on AA-Briefcase v1.1.[35] Artificial Analysis scored Grok 4.7 (xhigh) 46 on version 4.3.2 of its Intelligence Index, two points above Grok 4.6 (high), which scores 44 on that version. That 44 is not comparable with the 61 Grok 4.6 received on v4.1.1 at launch.[37][38][14]

Grok 4.6 stayed in the API. When checked on September 23, 2026, SpaceXAI's pricing page still listed grok-4.6 at its launch rates, its model detail page was online, and grok-4.6 and grok-4.7 were the only two models served on SpaceXAI's US regional endpoint.[39][40][41] The general models page had moved its recommendation to Grok 4.7 for code and chat, and the Grok 4.6 overview page in the documentation now redirects to the Grok 4.7 page.[42][4] The pricing page and release notes carried no grok-4.8 identifier on that date, so Musk's Grok 4.8 statements above still had no company documentation behind them.[39][36]

## See also

- [Grok](https://aiwiki.ai/wiki/grok)
- [Grok 4.5](https://aiwiki.ai/wiki/grok_4_5)
- [Grok 4.7](https://aiwiki.ai/wiki/grok_4_7)
- [Grok Build](https://aiwiki.ai/wiki/grok_build)
- [xAI](https://aiwiki.ai/wiki/xai)
- [Cursor](https://aiwiki.ai/wiki/cursor)
- [SpaceX-xAI merger](https://aiwiki.ai/wiki/spacex_xai_merger)
- [Artificial Analysis](https://aiwiki.ai/wiki/artificial_analysis)
- [SpaceX Starmind](https://aiwiki.ai/wiki/spacex_starmind)

## References

1. SpaceXAI. "Introducing Grok 4.6." August 12, 2026. https://x.ai/news/grok-4-6
2. SpaceXAI. "Introducing Grok 4.6." Archived copy of the launch-day page, captured August 12, 2026, 15:44 UTC. https://web.archive.org/web/20260812154457/https://x.ai/news/grok-4-6
3. SpaceXAI. "Release Notes." SpaceXAI Docs. Accessed August 12, 2026. https://docs.x.ai/developers/release-notes
4. SpaceXAI. "Grok 4.6." SpaceXAI Docs. Archived copy captured August 12, 2026, 15:47 UTC. https://web.archive.org/web/20260812154736/https://docs.x.ai/developers/grok-4-6 (the live URL, https://docs.x.ai/developers/grok-4-6, redirected to the Grok 4.7 page when checked on September 23, 2026)
5. SpaceXAI. "Models." SpaceXAI Docs. Accessed August 12, 2026. https://docs.x.ai/developers/models
6. SpaceXAI. "Pricing." SpaceXAI Docs. Accessed August 12, 2026. https://docs.x.ai/developers/pricing
7. SpaceXAI. "Reasoning." SpaceXAI Docs. Accessed August 12, 2026. https://docs.x.ai/developers/model-capabilities/text/reasoning
8. Cursor. "Introducing Grok 4.6." August 12, 2026. https://cursor.com/blog/grok-4-6
9. SpaceXAI (@SpaceXAI). "Introducing Grok 4.6. It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price." X, August 12, 2026. https://x.com/SpaceXAI/status/2087562800982077492
10. Elon Musk (@elonmusk). "Grok 4.5 and Opus 5 are alone on Pareto frontier" and reply "Grok 4.6 in 2 weeks and Grok 4.7 in 4 weeks." X, July 24, 2026. https://x.com/elonmusk/status/2080724087593226311
11. Elon Musk (@elonmusk). "Grok 4.6 releases around August 7. This will be the 1.5T model with significantly improved SFT & RL. Grok 4.7 will be the 2.1T model released a few weeks later..." X, July 28, 2026. https://x.com/elonmusk/status/2082123925283041545
12. Not a Tesla App. "Highlights From SpaceX's First-Ever Earnings Call: Starship, Starlink, Grok and More." August 2026. https://www.notateslaapp.com/news/4543/highlights-from-spacexs-first-ever-earnings-call-starship-starlink-grok-and-more
13. Tesla Oracle. "Space, Connectivity, AI: Key takeaways from Elon Musk's SpaceX Q2 2026 Earnings Call." August 10, 2026. https://www.teslaoracle.com/2026/08/10/space-connectivity-ai-key-takeaways-from-elon-musks-spacex-q2-2026-earnings-call/
14. Artificial Analysis. "Grok 4.6 (high): Intelligence, Performance and Price Analysis." Accessed August 12, 2026. https://artificialanalysis.ai/models/grok-4-6
15. Artificial Analysis. "LLM Leaderboard." Accessed August 12, 2026. https://artificialanalysis.ai/leaderboards/models
16. Arena AI. "Leaderboard." Accessed August 12, 2026. https://arena.ai/leaderboard
17. OpenRouter. "SpaceXAI: Grok 4.6." Model listing, accessed August 12, 2026. https://openrouter.ai/x-ai/grok-4.6
18. Reuters (via TradingView News). "SpaceXAI Introduces Grok 4.6." August 12, 2026. https://www.tradingview.com/news/reuters.com,2026:newsml_FWN449150:0-spacexai-introduces-grok-4-6/
19. Vals AI. "Harvey's Legal Agent Benchmark." Accessed August 12, 2026. https://www.vals.ai/benchmarks/hlab
20. TradingKey. "SpaceXAI Officially Launches Grok 4.6: Performance on AI Intelligence Analysis Index Matches GPT-5.6 Sol." August 12, 2026. https://www.tradingkey.com/analysis/stocks/us-stocks/262099926-space-xai-grok-4-6-matches-gpt-5-6-sol-ai-index-tradingkey
21. Elon Musk (@elonmusk). "SpaceX's massive corpus of world-class engineering data (excluding material blocked by ITAR) will be added during supplemental training of the 2T run." X, July 21, 2026. https://x.com/elonmusk/status/2079446276299465185
22. Elon Musk (@elonmusk). "Grok 4.7 will exceed all current models... the SpaceX training corpus is so awesome & unique that I would be shocked if any model is better at real-world engineering than 4.7." Reply to @cognition, X, August 12, 2026 (post since deleted; shown by X as deleted by its author when checked on September 23, 2026). https://x.com/elonmusk/status/2087606260539777263
23. Elon Musk (@elonmusk). "Grok 4.7 comes out in 10 days." X, September 2, 2026. https://x.com/elonmusk/status/2094983639780204846
24. Elon Musk (@elonmusk). "Grok 4.7 needs a few more days to cook." Reply to @farzyness, X, September 11, 2026. https://x.com/elonmusk/status/2098462085973741960
25. Elon Musk (@elonmusk). "Grok 4.7 should be roughly on par with Opus 5.0, not 5.1... Grok 4.9 is probably Astra/Fable class. Grok 5 maybe better than anything. We shall see." Reply to @itslueul, X, September 14, 2026. https://x.com/elonmusk/status/2099458047408013751
26. Tech Dev Notes (@techdevnotes). "Apart from Grok 4.7, what are we even expecting from SpaceXAI in september." X, September 13, 2026. https://x.com/techdevnotes/status/2099124531172434084
27. Elon Musk (@elonmusk). "Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL." Reply to @techdevnotes, X, September 14, 2026. https://x.com/elonmusk/status/2099308197802631191
28. Adam Lowisz (@AdamLowisz). "Elon, how close do you think Grok 4.8 will be towards achieving AGI?" X, September 14, 2026. https://x.com/AdamLowisz/status/2099308480372875729
29. Elon Musk (@elonmusk). "That will be Grok 5." Reply to @AdamLowisz, X, September 14, 2026. https://x.com/elonmusk/status/2099455592670634034
30. Elon Musk (@elonmusk). "Truly massive gains will come in ~3 months when the entire training and inference stack is written in C/C++ and massively simplified... and we exact-map Grok to work incredibly well on a GB300." Reply to @aaronburnett, X, June 29, 2026. https://x.com/elonmusk/status/2071385784154759468
31. Elon Musk (@elonmusk). "Our training (aka pre-training) software is now the internally developed C/C++ stack, which was written primarily by the Starlink software team... However, our reinforcement learning stack is still somewhat of a legacy pile of many different third-party software systems." Reply to @aaronburnett, X, September 14, 2026. https://x.com/elonmusk/status/2099500295931789493
32. Elon Musk (@elonmusk). "Our C/C++ training stack was written by humans. AI is not yet good enough for extremely high performance software (but it will be)." Reply to @PatrickToulme, X, September 14, 2026. https://x.com/elonmusk/status/2099542714974896467
33. SpaceXAI. "Models." SpaceXAI Docs. Accessed September 15, 2026. https://docs.x.ai/developers/models
34. SpaceXAI. "Release Notes." SpaceXAI Docs. Accessed September 15, 2026. https://docs.x.ai/developers/release-notes
35. SpaceXAI. "Introducing Grok 4.7." September 21, 2026. Accessed September 23, 2026. https://x.ai/news/grok-4-7
36. SpaceXAI. "Release Notes." SpaceXAI Docs. Accessed September 23, 2026. https://docs.x.ai/developers/release-notes
37. Artificial Analysis. "Grok 4.7 (xhigh): Intelligence, Performance and Price Analysis." Accessed September 23, 2026. https://artificialanalysis.ai/models/grok-4-7
38. Artificial Analysis. "Grok 4.6 (high): Intelligence, Performance and Price Analysis." Accessed September 23, 2026. https://artificialanalysis.ai/models/grok-4-6
39. SpaceXAI. "Pricing." SpaceXAI Docs. Accessed September 23, 2026. https://docs.x.ai/developers/pricing
40. SpaceXAI. "grok-4.6" model detail page. SpaceXAI Docs. Accessed September 23, 2026. https://docs.x.ai/developers/models/grok-4.6
41. SpaceXAI. "Regional Endpoints." SpaceXAI Docs. Accessed September 23, 2026. https://docs.x.ai/developers/advanced-api-usage/regions
42. SpaceXAI. "Models." SpaceXAI Docs. Accessed September 23, 2026. https://docs.x.ai/developers/models