Grok vs Claude: plans, API prices and benchmarks compared

Grok vs Claude comes down to price, models and published results. As of October 28, 2026, Claude's paid plans start at $20 a month ($17 billed annually) and Grok's start at $8 a month through xAI's X Subscription, and on Artificial Analysis's Intelligence Index the top Claude model scores 58 against 35 for the best Grok 4.7 setting. This comparison is based on the products' docs and published results, not on our own hands-on testing.

Grok is the cheaper way in and Claude has the higher published scores. The details below explain what each plan gets you and where the numbers come from.

Grok vs Claude plans and prices

Prices are from Claude's pricing page and xAI's X Subscription pricing page, both read on October 28, 2026. Prices exclude sales tax.

The two companies measure usage differently. Claude says its limits run on a rolling five-hour window, with weekly limits on paid plans, and gives no fixed message count. xAI sells multiples of an undefined "base" pool that Grok, Grok Bot and Cursor share. Neither page tells you how many messages you get.

For the free tiers in detail, see our articles on whether Claude is free and whether Grok is free.

Models and API prices

Claude's pricing page lists Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5. xAI's models page recommends Grok 4.7 for code and chat. All prices below are per million tokens, read on October 28, 2026.

Grok 4.7 matches Sonnet 5.5 on input price and costs less on output. Claude has the cheaper small model and the pricier top tier. Claude's page lists a 50% discount for batch processing.

What the independent benchmarks show

Artificial Analysis publishes an Intelligence Index that combines many tests, with cost per task and speed. We read its leaderboard on October 28, 2026. The page shows no update date, and its homepage mentions Intelligence Index v5. The scores below are the highest and the most comparable settings for each model.

At their higher settings, Opus 5.5, Sonnet 5.5 and Fable 5.1 all score well above the best Grok 4.7 setting. Opus 5.5 at high effort scores 52 for $3.46 a task, against 35 for $6.18 on Grok 4.7 at high. Haiku 5.5 at max scores about the same as Grok 4.7 and costs far less per task, $0.41, though the page does not explain how the cost is measured.

On the site's domain capability table, Claude Opus 5.5 ranks first in all six areas, including finance, legal and healthcare, and Grok 4.7 ranks between ninth and twelfth. That table is a snapshot of selected models.

One caution. The Claude entries at the highest settings show very long times to first response, 700 seconds for Opus 5.5 at max, so the top scores come with a wait. And these are one site's tests, not every task you care about. Other leaderboards such as SWE-bench and Epoch AI test different things.

Which one fits you

Our read, from the docs and the numbers above: if you want the highest published scores, or you code and need Claude Code on a $20 plan, Claude is the stronger choice. If you already pay for X, want a lower entry price or want Grok bundled with Cursor, Grok's $8 Plus plan costs less than Claude Pro, though it has no Grok Bot and no stated usage size.

Both free tiers are worth trying before you pay, since neither company publishes a message count for them.

What we don't know

xAI's pages do not say which models consumer Grok plans use, and neither company gives message counts. Claude's page does not list the Max 20x price separately. Plans, limits and benchmark scores change, so check the pages before you pay. For another matchup, see ChatGPT vs Grok.