Grok
xAIGrok 4.7 is GA on Microsoft Foundry Global and US Data Zone; Grok 4.6 remains a Global Standard preview.
xAI released Grok 4.7 on 2026-09-21 as its current direct-API flagship: 500K context, $2/$0.50/$6 per M for input/cached input/output, and $4/$1/$12 once a prompt reaches the published 200K long-context threshold. The separate fast variant is announced by xAI but is not represented as a second token lane here.
Microsoft Foundry made Grok 4.7 generally available on October 6 in Global Standard and US Data Zone Standard, with a 500K context window. Foundry's rates follow the same short- and long-context schedule as xAI Direct, with the documented 1.1x US Data Zone premium. Azure Retail Prices now lists Grok 4.7 token meters effective October 1, confirming these lanes; it also publishes separate EU and US Government Data Zone rates whose deployment support is not stated in the launch article.
Grok 4.6 (500K context on the direct API), released 2026-08-12, remains a public-preview Global Standard deployment in Foundry. Microsoft documents a 200K-token context window and 128K maximum output for that preview; its separate 4.6 Data Zone retail meters remain unmodeled because Microsoft has not confirmed that deployment path.
Pricing
per 1M tokens · USD / CHF| Model | Tier / Host | Context | Input | Cached | Output | Confidence |
|---|---|---|---|---|---|---|
Grok 4.7 xAI's current flagship for coding and knowledge work, released 2026-09-21. Standard rates are $2/$0.50/$6 per M; requests at or above 200K prompt tokens use the separate long-context lane below. xAI also announces a faster variant at twice the price, but does not expose it as a separate token table on the official pricing page. | Direct API xAI direct API | 500K | $2.00CHF 1.61 | $0.50CHF 0.403 | $6.00CHF 4.83 | official xAI's official Grok 4.7 announcement (x.ai/news/grok-4-7), published and captured 2026-09-21, states that the model is available through the Grok API from $2/M input and $6/M output, with $0.50/M cached input and a 500K context window. The official model page (docs.x.ai/developers/models/grok-4.7) confirms the same model identity, context and cache rate; xAI's pricing page (docs.x.ai/developers/pricing) publishes the $4/$1/$12 long-context band for requests at or above 200K prompt tokens. Microsoft's October 6 Foundry launch separately publishes the Global Standard and US Data Zone rates; the Azure Retail Prices feed captured 2026-10-10 now lists the corresponding Foundry meters effective 2026-10-01. |
Grok 4.7 Long Context Long-context pricing band for requests at or above 200K prompt tokens. This is a separate comparison lane so the site does not confuse a workload's cumulative monthly input volume with one prompt's context length. | Direct API xAI direct API | 500K | $4.00CHF 3.22 | $1.00CHF 0.805 | $12.00CHF 9.66 | official xAI's official pricing page (docs.x.ai/developers/pricing), captured 2026-09-21, publishes Grok 4.7 at $4/M input, $1/M cached input and $12/M output for the long-context band beginning at 200K prompt tokens. The model page (docs.x.ai/developers/models/grok-4.7) confirms the 500K context window. Microsoft's October 6 Foundry launch separately publishes Global Standard and US Data Zone rates; the Azure Retail Prices feed captured 2026-10-10 now lists the corresponding Foundry meters effective 2026-10-01. |
Grok 4.7 Microsoft Foundry Global Standard, generally available from October 6. Prompts under 200K tokens use the short-context rates; longer prompts use the separate lane below. | Foundry ·Global | 500K | $2.00CHF 1.61 | $0.50CHF 0.403 | $6.00CHF 4.83 | official Microsoft's Foundry blog, published 2026-10-06 and captured 2026-10-09, lists Grok 4.7 Global Standard at $2/M input, $0.50/M cached input and $6/M output for prompts under 200K tokens, with a 500K context window. The Azure Retail Prices Foundry feed captured 2026-10-10 lists the matching Global input, cached-input and output meters effective 2026-10-01. |
Grok 4.7 Long Context Microsoft Foundry Global Standard long-context rates for prompts of 200K tokens or more. | Foundry ·Global | 500K | $4.00CHF 3.22 | $1.00CHF 0.805 | $12.00CHF 9.66 | official Microsoft's Foundry blog, published 2026-10-06 and captured 2026-10-09, lists Grok 4.7 Global Standard at $4/M input, $1/M cached input and $12/M output for prompts of 200K tokens or more, with a 500K context window. The Azure Retail Prices Foundry feed captured 2026-10-10 lists the matching long-context Global input, cached-input and output meters effective 2026-10-01. |
Grok 4.7 Microsoft Foundry US Data Zone Standard, generally available from October 6. The listed rates are 1.1x Global Standard; longer prompts use the separate lane below. | Foundry ·Data Zone | 500K | $2.20CHF 1.77 | $0.55CHF 0.443 | $6.60CHF 5.31 | official Microsoft's Foundry blog, published 2026-10-06 and captured 2026-10-09, lists Grok 4.7 US Data Zone Standard at $2.20/M input, $0.55/M cached input and $6.60/M output for prompts under 200K tokens. The Azure Retail Prices Foundry feed captured 2026-10-10 lists the matching commercial-US Data Zone meters effective 2026-10-01. It also publishes separate EU and US Government Data Zone rates; see the provider note because those deployment combinations are not stated in Microsoft's launch article. |
Grok 4.7 Long Context Microsoft Foundry US Data Zone Standard long-context rates for prompts of 200K tokens or more; 1.1x the Global Standard rate. | Foundry ·Data Zone | 500K | $4.40CHF 3.54 | $1.10CHF 0.886 | $13.20CHF 10.63 | official Microsoft's Foundry blog, published 2026-10-06 and captured 2026-10-09, lists Grok 4.7 US Data Zone Standard at $4.40/M input, $1.10/M cached input and $13.20/M output for prompts of 200K tokens or more. The Azure Retail Prices Foundry feed captured 2026-10-10 lists the matching long-context Data Zone input, cached-input and output meters effective 2026-10-01. |
Grok 4.6 xAI's new flagship, released 2026-08-12 and labeled 'Latest' in xAI's docs, superseding Grok 4.5 as the headline model. 500K context; text+image input, text output. All rates double for requests with ≥200K prompt tokens ($4 input / $1 cached / $12 output). Cached input is higher than Grok 4.5's ($0.50 vs $0.30); input and output prices are identical. | Direct API xAI direct API | 500K | $2.00CHF 1.61 | $0.50CHF 0.403 | $6.00CHF 4.83 | official xAI's official pricing page (docs.x.ai/developers/pricing): $2/M input, $0.50/M cached input, $6/M output, 500K context, doubling to $4/$1/$12 for prompts ≥200K tokens. Release confirmed by docs.x.ai/developers/release-notes, entry dated August 12, 2026. Both pages were read via raw DOM on 2026-08-12; the Azure sweep on that date had no 4.6 meter, before Foundry publication on the effective 2026-09-01 tranche. |
Grok 4.6 Microsoft Foundry public preview, Global Standard only. The Foundry preview documents a 200K context window, distinct from xAI's 500K direct-API window; the current Foundry Global retail meters match xAI's short-context token rates. | Foundry ·Global | 200K 128K max out | $2.00CHF 1.61 | $0.50CHF 0.403 | $6.00CHF 4.83 | official Azure Retail Prices API, serviceName 'Foundry Models', productName 'Azure Grok Models', captured 2026-09-11: effective 2026-09-01 meters '4.6 Inp Glbl Tokens' $0.002/1K ($2.00/M), '4.6 Cached Glbl Tokens' $0.0005/1K ($0.50/M), and '4.6 Outp Glbl Tokens' $0.006/1K ($6.00/M), with the commercial majority across 39 regions; the two US Government regions carry a 1.25x outlier and are excluded. Microsoft's current Foundry deployment guide documents Grok 4.6 as preview and Global Standard only, so the separate 4.6 Data Zone meter group is not inferred into this catalog. |
Grok 4.6 Long Context Foundry's long-context pricing applies once a prompt reaches 200K input tokens; the preview's documented context window remains 200K. This is a pricing band, not a claim of a larger Foundry context window. | Foundry ·Global | 200K 128K max out | $4.00CHF 3.22 | $1.00CHF 0.805 | $12.00CHF 9.66 | official Azure Retail Prices API, serviceName 'Foundry Models', productName 'Azure Grok Models', captured 2026-09-11: effective 2026-09-01 meters '4.6 Inp Glbl L Tokens' $0.004/1K ($4.00/M), '4.6 Cached Glbl L Tokens' $0.001/1K ($1.00/M), and '4.6 Outp Glbl L Tokens' $0.012/1K ($12.00/M), with the commercial majority across 39 regions; the two US Government regions carry a 1.25x outlier and are excluded. The long-context threshold and direct-API schedule are corroborated by xAI's official pricing page. |
Grok 4.5 Flagship. All rates double for requests with ≥200K prompt tokens ($4 input / $12 output). | Direct API xAI direct API | 500K | $2.00CHF 1.61 | $0.30CHF 0.241 | $6.00CHF 4.83 | official xAI's official Grok 4.5 model page: $2/M input, $0.30/M cached input, $6/M output, and 500K context; captured 2026-07-21. |
Grok-4.3 Newest Grok on Foundry; cached-input meter now published. Long-context (≥200K prompt) meters bill at 2x, including cached input at $0.40/M. | Foundry ·Global | — | $1.25CHF 1.01 | $0.20CHF 0.161 | $2.50CHF 2.01 | official Input $1.25/M and output $2.50/M from the Foundry Models pricing page. Cached input from the Azure Retail Prices API meter named bare '4.3' — a 'grok' meterName search misses it — 'Cached Inp Glbl' at $0.0002/1K ($0.20/M), effective 2026-05-01, captured 2026-07-23; matches xAI's direct Grok 4.3 docs. Long-context 'L' meters exist at 2x ($2.50 in / $5.00 out / $0.40 cached). |
Grok-4.3 Data Zone premium: exactly 1.1x the Global rate on all three meters. Long-context (≥200K prompt) DZ meters also exist, billing at $2.75 in / $5.50 out / $0.44 cached per M. | Foundry ·Data Zone | — | $1.38CHF 1.11 | $0.22CHF 0.177 | $2.75CHF 2.21 | official Azure Retail Prices API '4.3 Inp/Cached Inp/Outp DZ' meters, effective 2026-05-01, captured 2026-07-26: $1.375/M input, $0.22/M cached input, $2.75/M output — a clean 1.1x the tracked Global rate ($1.25/$0.20/$2.50). Long-context 'L' DZ meters re-verified 2026-07-28: '4.3 Inp DZ L Tokens' $0.00275/1K ($2.75/M), '4.3 Cached Inp DZ L Tokens' $0.00044/1K ($0.44/M), '4.3 Outp DZ L Tokens' $0.0055/1K ($5.50/M) — matching the Global long-context row's 2x pattern at the same 1.1x DZ premium. |
Grok 4.1 Fast Cheap fast tier on Foundry. | Foundry ·Global | — | $0.20CHF 0.161 | — | $0.50CHF 0.403 | official Microsoft Foundry Models pricing page (Grok tab), captured 2026-07-19. |
Quirks & gotchas
Grok 4.7 Data Zone rates vary by region
The Azure Retail Prices feed lists Grok 4.7 Data Zone meters effective 2026-10-01 at $2.20/$0.55/$6.60 per M (short context) and $4.40/$1.10/$13.20 (long context) for commercial US regions. It also lists EU Data Zone rates of $2.40/$0.60/$7.20 and $4.80/$1.20/$14.40, plus US Government Data Zone rates of $2.75/$0.6875/$8.25 and $5.50/$1.375/$16.50. Microsoft's launch article confirms Global and commercial US Data Zone deployments only, so the comparison keeps those documented lanes while the other region prices await deployment confirmation.
Grok 4.6 Foundry is Global-only for now
Microsoft's current deployment guide lists Grok 4.6 as a public-preview Global Standard model with a 200K context window and 128K maximum output. The Azure Retail Prices feed also emits 4.6 Data Zone meters, but because the current model guide does not list a Data Zone deployment, those meters are deliberately left unmodeled until Microsoft confirms that deployment path.
Grok 4.5 remains Direct-only
Use xAI's Direct API for Grok 4.5. Microsoft Foundry now lists Grok 4.7 in Global and US Data Zone Standard and Grok 4.6 in Global Standard preview; do not confuse those lanes with availability of the 4.5 model.
Grok-4.3 now has a Foundry cache meter
The Azure Retail Prices API now publishes a cached-input meter for Grok-4.3 at $0.20 / CHF 0.16 per M (effective 2026-05-01, surfaced on the 2026-07-23 recheck), matching xAI's direct Grok 4.3 cache rate. The meter is named bare '4.3', so a 'grok' search misses it. Grok 4.1 Fast still shows no cached column on Foundry.
Long-context surcharge on the direct API
On xAI's API, requests at or above 200K prompt tokens double every meter: $4 / CHF 3.22 input and $12 / CHF 9.66 output per M. The 500K window is usable, but the second half of it bills at 2x.