--- title: Grok 4.7 lands at $6 output while matching Opus-class... description: Grok 4.7 launched on Vercel AI Gateway at $6 per million output tokens, priced well below Opus/Sonnet-class competitors while reportedly matching their... canonical: https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Grok 4.7 lands at $6 output while matching Opus-class performance | daily.dev og:description: Grok 4.7 launched on Vercel AI Gateway at $6 per million output tokens, priced well below Opus/Sonnet-class competitors while reportedly matching their... og:url: https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw og:image: https://api.daily.dev/og/posts/1KukJBDBw.png og:image:alt: Grok 4.7 lands at $6 output while matching Opus-class performance og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Grok 4.7 lands at $6 output while matching Opus-class performance **[Trends](https://daily.dev/sources/trends)** · 2 min read · 5 upvotes · 1 comments ## Summary Grok 4.7 launched on Vercel AI Gateway at $6 per million output tokens, priced well below Opus/Sonnet-class competitors while reportedly matching their performance tier. Elon Musk described it as roughly on par with Opus 5.0, and analyst estimates put it around 2 trillion parameters. It offers a 500K token context window, adjustable reasoning levels (low, medium, high, xhigh), and works through the AI SDK, an OpenAI-compatible API, and coding agents like Cursor, Codex, and Amp. Vercel is running a 40% discount through September 27, and the model reportedly leaked earlier via GCP quota pages before its official release. ## Content xAI dropped Grok 4.7 last week with Elon Musk promising it would "exceed all current models" and be more token-efficient than 4.6. The community has been stress-testing that claim ever since, and the results are not flattering. The core problem: Grok 4.7 is burning roughly double the tokens per task compared to 4.6 (around 81k vs 38k), which makes the "better token efficiency" claim hard to defend. Theo, who's been running benchmarks obsessively, put it bluntly: "I have yet to find a single bench where Grok 4.7 is more token-efficient than Grok 4.6." He also clocked it as more expensive than Astra on Artificial Analysis once you account for actual task cost, not just sticker price. On benchmarks, the picture is mixed at best. Grok 4.7 jumped from 20.3% to 38.0% on Terminal-Bench 4.0, which is a real gain. It ranks #4 on Artificial Analysis's Coding Agent Index. But it's still behind Claude Fable 5.1 (57.9% on Terminal-Bench), and the Intelligence Index barely moved, up just 2 points. One reviewer's KingBench 3 suite had it at 77.5%, tying Kimi K3 and Opus 5 but well behind Astra's 90%. SVG generation and front-end tasks were recurring weak spots. The pricing math is where things get interesting. At $6 per million output tokens, it's genuinely cheap compared to OpenAI ($20) and Anthropic ($50). Vercel is running a 40% discount through September 27th, and it's now in GitHub Copilot and v0. Some people are calling it a steal for backend agentic work. Others point out that when a model uses 2x the tokens, the cheap per-token rate stops mattering. The reception is split pretty cleanly. Kent C. Dodds and Matt Shumer both called it their new daily driver. ScriptedAlchemy ran it for 70+ hours and liked what they saw. Theo called it "a disappointing release" and documented a Fishslop benchmark run he described as "the worst I've seen this year." The consensus forming in the middle: solid for backend engineering and long agentic tasks, weak on front-end and visual work, and probably best used as a sub-agent rather than a lead model. Elon's pre-release framing that it would "exceed all current models" is the part that's aging badly. The model has real strengths. The marketing did it no favors. ## Questions this post answers ### What is the output token pricing for Grok 4.7 on Vercel AI Gateway? Grok 4.7 costs $6 per million output tokens on Vercel AI Gateway, undercutting comparable Opus/Sonnet-class model pricing from Anthropic. It ships with a 500K token context window and adjustable reasoning levels (low, medium, high, xhigh). Vercel is also running a 40% discount through September 27, further lowering the effective cost. _Comparing model pricing before locking in a coding agent stack is easier when tracked on daily.dev._ ### How does Grok 4.7 compare to Claude Opus 5.0 in performance? Elon Musk stated Grok 4.7 should be roughly on par with Opus 5.0, not Opus 5.1, calling it better in some ways and worse in others. Analyst @scaling01 places it alongside Opus and Sol in capability, estimating around 2 trillion parameters, though real-world production testing is still needed to confirm parity, especially for complex reasoning at the xhigh setting. _Developers weighing Grok against Opus for production workloads can follow benchmark discussions on daily.dev._ ### Which coding agents support Grok 4.7 through Vercel AI Gateway? Grok 4.7 works across the Vercel AI SDK, an OpenAI-compatible Chat Completions API, and coding agents including Cursor, Codex, and Amp. It's called via the model identifier spacexai/grok-4.7 and comes with no markup, no platform fee, Zero Data Retention, and no prompt training on submitted data. _Teams picking a coding agent and model combo can compare integration options on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 30 comments across x (as of 2026-09-24). **TL;DR:** Reactions to Grok 4.7 are largely underwhelmed, with several people saying the release feels disappointing and outshined by other model launches happening the same week. **Sentiment:** 10% positive · 25% mixed · 65% skeptical **The case for** - A few found it strong for code review reasoning, even if not for writing code. **The pushback** - Multiple people report the release feels disappointing compared to expectations. - Concerns that it's slower and burns more tokens, which matters more than a small benchmark bump. - Some feel newer 'smarter' benchmarked models don't actually feel better in daily use. - Attention is being pulled toward other simultaneous model releases, overshadowing this one. **By community** - x (skeptical): Replies skew disappointed, with people saying they already forgot about the release and are more focused on competing model launches. **Hottest debate:** Whether the benchmark improvement is even meaningful if it feels slower and more token-hungry in practice. **Open questions** - Will a follow-up patch release address the reported regressions? - Why release this now when attention is on competing models? **Highlights** > @theo AI models are hitting a weird point where “smarter” doesn’t always feel better. If users prefer the older model, the benchmark doesn’t matter > — [Dmytromakes on x](https://x.com/Dmytromakes/status/2102495998199107711) > @theo if it gets slower and burns more tokens, a small benchmark bump won’t save it. speed matters when you’re using it all day. > — [jasperwu\_ on x](https://x.com/jasperwu_/status/2102497398253871483) > @theo Wait until you see the insane fucking level of regression in both 6 Sol and Opus 5.5. It’s genuinely so disappointing > — [yzycrave on x · 1 points](https://x.com/yzycrave/status/2102496975530643608) > @theo I agree I'm quite disappointed too. > — [cjtrade01 on x](https://x.com/cjtrade01/status/2102494780286464347) > @theo Calling the release disappointing in public is the rare AI launch ritual that might actually get the next one fixed. > — [OffByOne67 on x](https://x.com/OffByOne67/status/2102495671236637043) **Source threads** - [x](https://x.com/theo/status/2102493204641939663) · 0 points · 30 comments ## Community discussion Top comments from developers on daily.dev. **@amizzo** · 1 upvotes > "Elon’s pre-release claim that it would “exceed all current models” is not what the numbers show." > > I AM. SHOCKED! --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#grok](https://daily.dev/tags/grok) [View this post on daily.dev](https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Grok 4.7 lands at $6 output while matching Opus-class performance","url":"https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw"},"datePublished":"2026-09-21T16:10:17.398Z","dateModified":"2026-09-24T16:23:31.471Z","description":"Grok 4.7 launched on Vercel AI Gateway at $6 per million output tokens, priced well below Opus/Sonnet-class competitors while reportedly matching their...","image":"https://pbs.twimg.com/media/HSwLTt6XwAATndb.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HSwLTt6XwAATndb.jpg","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"llm,ai-agents,grok","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Grok 4.7 lands at $6 output while matching Opus-class performance"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw","comment":[{"@type":"Comment","text":"“Elon’s pre-release claim that it would “exceed all current models” is not what the numbers show.”\nI AM. SHOCKED!","datePublished":"2026-09-22T03:49:27.657Z","url":"https://daily.dev/posts/1KukJBDBw#c-nxsRGbi8o","author":{"@type":"Person","name":"Anthony Izzo","url":"https://daily.dev/amizzo","image":"https://lh3.googleusercontent.com/a/ACg8ocI3KpxrCJphZmqCBYWzrlZz0yUJS3pQF3RYarylAe8SP_G6_KUCoQ=s96-c"},"interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/grok-4-7-lands-at-6-output-while-matching-opus-class-performance-1kukjbdbw#faq","mainEntity":[{"@type":"Question","name":"What is the output token pricing for Grok 4.7 on Vercel AI Gateway?","acceptedAnswer":{"@type":"Answer","text":"Grok 4.7 costs $6 per million output tokens on Vercel AI Gateway, undercutting comparable Opus/Sonnet-class model pricing from Anthropic. It ships with a 500K token context window and adjustable reasoning levels (low, medium, high, xhigh). Vercel is also running a 40% discount through September 27, further lowering the effective cost. Comparing model pricing before locking in a coding agent stack is easier when tracked on daily.dev."}},{"@type":"Question","name":"How does Grok 4.7 compare to Claude Opus 5.0 in performance?","acceptedAnswer":{"@type":"Answer","text":"Elon Musk stated Grok 4.7 should be roughly on par with Opus 5.0, not Opus 5.1, calling it better in some ways and worse in others. Analyst @scaling01 places it alongside Opus and Sol in capability, estimating around 2 trillion parameters, though real-world production testing is still needed to confirm parity, especially for complex reasoning at the xhigh setting. Developers weighing Grok against Opus for production workloads can follow benchmark discussions on daily.dev."}},{"@type":"Question","name":"Which coding agents support Grok 4.7 through Vercel AI Gateway?","acceptedAnswer":{"@type":"Answer","text":"Grok 4.7 works across the Vercel AI SDK, an OpenAI-compatible Chat Completions API, and coding agents including Cursor, Codex, and Amp. It's called via the model identifier spacexai/grok-4.7 and comes with no markup, no platform fee, Zero Data Retention, and no prompt training on submitted data. Teams picking a coding agent and model combo can compare integration options on daily.dev."}}]} ```