Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

.@Grok in your Tesla can now do meaningful work for you With Connectors, you can manage your inbox, clean up your calendar, or talk through existing files/chat/tasks – all hands-free

Grok 4.7 is behind only Anthropic models on AA-Briefcase, ranking just behind Opus 5 at ~50% of its Cost per Task Grok 4.7’s improvements over Grok 4.6 are clear in AA-Briefcase-Lite, our public due diligence scenario where models are tasked with building market models and target assessment decks. Grok 4.7 gains significantly in Analytical Quality Elo (1698 → 1994) with a slight regression in Presentation Elo (1531 → 1499). API cost to produce example decks: Grok 4.7 (xhigh) ~$8 vs. Grok 4.6 (xhigh) ~$4.40

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 by @SpaceXAI and @elonmusk is now in the Agent Arena! Your votes drive the @arena leaderboards, head over and bring your toughest prompts. In Agent Arena, we measure models on millions of real-world, long-horizon agentic tasks from a global community of users. Models can access web search, filesystem, and terminal tools to complete complex workflows. The leaderboard measures model performance on outcomes relative to the average model using a causal tracing methodology. In addition to Agent Arena, Grok 4.7 is in Battle Mode for: Text, Vision, Code, and Document.

Grok 4.7 scores 46 on the Artificial Analysis Intelligence Index to bring SpaceXAI into the top 4 AI labs. Coding Agent Index performance has also improved, overtaking GPT-5.6 Sol Grok 4.7 scores +2 points over Grok 4.6 on the Intelligence Index, with strong performance on agentic knowledge work tasks. We evaluated the new model at xhigh reasoning effort. Congratulations to @SpaceXAI and @ElonMusk on the release! Key takeaways: ➤ Grok 4.7 joins the frontier of agentic knowledge work: Grok 4.7 gains +111 Elo over Grok 4.6 (high) on AA-Briefcase, our private benchmark for long-horizon agentic knowledge work, scoring 1657 Elo and placing it alongside Claude Opus 5 and Claude Fable 5.1 at the frontier. On GDPval-AA, it scores 1695 Elo, +90 ahead of Grok 4.6 (high). ➤ A leap in coding agent performance: Grok 4.7 (xhigh) with Grok Build scores 56 on the Artificial Analysis Coding Agent Index, up +9 points from Grok 4.6 (xhigh). Among models in their native harnesses, Grok 4.7 + Grok Build now ranks 4th, behind only Claude Fable 5.1, GPT-6 Astra, and Claude Opus 5. ➤ Incremental performance changes elsewhere: Outside of agentic knowledge work, Grok 4.7 broadly matches Grok 4.6 (high) on the other Intelligence Index tasks. It improves on Terminal-Bench 4.0 (+4.5 percentage points) and GDP.pdf (+3.0 p.p.), with regressions on AA-LCR (-3.7 p.p.) and AutomationBench-AA (-1.1 p.p.). ➤ High token use across tasks: Grok 4.7's gains come with higher token usage. Grok 4.7 (xhigh) uses approximately 81k output tokens per Intelligence Index task, compared with 36k for Grok 4.6 (high) and 27k for GPT-6 Astra (max) - 125% and 196% more, respectively. Other model details: ➤ Context window of 500k tokens, unchanged from Grok 4.6 ➤ Pricing of $2/$6 per 1M input/output tokens with cache hits discounted to $0.50 per 1M tokens, matching Grok 4.6 ➤ Configurable reasoning effort spans low to xhigh. Our evaluation uses xhigh.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

@farzyness Grok 4.7 needs a few more days to cook. We might have penalized response length too much (or something) in RL, as it still gives up on hard tasks (that it can do!) too early and isn’t yet sufficiently rigorous in checking its work.

Grok Bot Summary of SpaceX CFO Bret Johnsen at Goldman Sachs Communacopia today. Vertical integration Vertical integration is the company’s core operating model, not a side strategy. - Rockets: own metal → engines → avionics → software - Starlink: own launch, satellites, and the end customer - AI: build facilities and power themselves, run their own models, sell to consumer and enterprise, and soon orbital compute Starship and launch Starship is the foundation for every other business. - Flight 13: big learning flight. Delivered demo V3 payloads, relit a Raptor, and got a soft, precise second-stage splashdown. Recovery team towed the stage back so engineers could study the heat shield. - Those learnings feed straight into Flight 14 and beyond. - Flight 14 (later this month): first revenue-generating Starship flight, flying production V3 Starlink satellites. - Later this year: aim to recover both first and second stages. Orbital compute Most of the AI industry agrees orbital compute is the future. Almost everyone else thinks it’s many years away. SpaceX disagrees because they control the stack. - Target: first orbital compute satellites next year - Scale: big compute in space into 2028 - Hardware approach: same V3 bus as Starlink, swap the payload, add larger solar arrays Why orbital can beat terrestrial on cost The crossover is about Starship reusability. - Falcon 9: first-stage reuse since Dec 2015; 500+ booster reflights - Starship: first stage already recovered/reflown; second-stage recovery progressing - Goal: reflight of both stages as soon as next year, which drops deployment cost sharply Terrestrial compute is getting more expensive (power, cooling, buildings, real estate). Orbital rides the opposite curve: cheaper rockets + better/cheaper satellites + scale. Johnsen said cost parity could come as soon as next year. Terrestrial compute and the $100B ARR goal - End of this year: on track for ~$100B ARR (annualizing the December number) - New update: another hosting deal closed earlier this month → about $1.1B/month starting Dec 1 → roughly +$13B ARR - Capacity: end this year well over 2 GW; next year 5–10 GW deployed - Confidence comes from line of sight to power, facilities, and permitting, plus being NVIDIA-exclusive for allocation - They stand compute up fast for themselves and for industry partners, which strengthens the NVIDIA relationship How they monetize compute Most hosting deals are short: ~90 days with a 90-day out (~6-month commits), including the newest deal. Why keep them short? - High conviction in their own products (Grok, Grok Bot, Cursor team after closing that deal) - Don’t want to lock forever capacity they may need internally - Internal bar: don’t let internal monetization fall below external hosting Earnings framing for next year: roughly $30–$50 per watt monetization range; they said they’re at the high end. Hosting customers appear to monetize even higher, which is why demand stays strong. Payback is under one year on new compute capex, so residual GPU value and financing options look attractive. “Not all CapEx is the same” — GPUs with <1-year payback are different from a launch tower built for decades. AI products and M&A Historically SpaceX was almost all organic growth. This year they did M&A because the AI product cycle rewards speed to frontier. - Closed Cursor deal weeks ago; product cycles already accelerating (called out Grok Bot) - Grok 4.6 improved on 4.5; 4.7 coming soon - Pitch: best infrastructure + competitive model + lower token cost = best position for customers - Market mood shift: months ago people bought the infra story but doubted the products; ~90 days later that skepticism is fading Starlink broadband Started as “better than nothing” (~2020–21). Now enterprise-grade with strong uptime/SLAs. - Resiliency pitch: boards will ask why Starlink wasn’t in the network if you go down - Mobility: aircraft backlog is large and production is ramping; cruise ships, yachts, trains too - Awareness, especially outside the US, is still a growth unlock - Longer-term: physical AI (robots, cars, aircraft) will need always-on connectivity terrestrial networks can’t fully cover Mobile / direct-to-cell Not a distraction. Same V3 bus, different payload. - Fly direct-to-device satellites through next year - Target service turn-on: first half of 2028 - V1 today (e.g. T-Mobile / T-SAT): text / light voice, great for emergencies and dead zones - Next gen: full 5G-quality from space - US: mid-band spectrum from EchoStar, FCC path for space + terrestrial - Go-to-market: flexible — own terrestrial build, or partner with carriers - International: same regulator-by-regulator playbook as broadband (Starlink now in 170+ countries) Near-term priorities: 1. Starship (enables everything else) 2. Terrestrial compute (funds growth and teaches them how to do orbital) Bottom line in one line Own the full stack, make Starship reusable at scale, use terrestrial AI compute as a cash engine now, and use the same satellite bus + Starship cadence to win broadband, mobile, and orbital AI.

JUST IN: El Salvador’s AI tutoring pilot in 171 public schools produced reading, math, &amp; science results above the national average &amp; comparable to Germany &amp; Sweden.

Grok Imagine Video 1.5 agent is now available. Powered by our newest Image 2.0 model, it delivers higher quality, better storytelling from a smarter agent and excels at connecting multiple shots together with greater continuity.

Grok Bot is now available for iPad

You can now add Bot templates from our marketplace. We’re sharing Haggle Bot, our in-house procurement specialist. It negotiates vendor contracts, finds unused SaaS seats, and price-checks recurring purchases. One week in, it's saved us over $100K.

Grok Bot for Enterprise is available today. It’s free for all Grok and Cursor enterprise customers for the next two weeks. https://x.ai/news/grok-bot-for-enterprise