Is GPT-5.6 Standard Global pricing currently 0.01 per 1M tokens?

Question

Is GPT-5.6 Standard Global pricing currently 0.01 per 1M tokens?

The Max Coder 5

I noticed that GPT-5.6 Sol usage is showing the same rate, 0.01 per 1M tokens, for cached input, cache writes, input, and output.

The same pattern appears for GPT-5.6 Luna and Terra. But GPT-5.5 usage in the same Azure export has different rates by token type, such as 0.50 cached input, 5.00 input, and 30.00 output.

Is the GPT-5.6 rate intentional, perhaps a temporary promotion, or could these new meters be incorrectly priced and re-rated later?

Thanmayi Godithi 11,550 Reputation points Microsoft External Staff Moderator

2026-07-17T14:47:14.22+00:00
Hi The Max Coder ,

Thank you for reporting this observation.

Based on currently published Azure pricing guidance, GPT-5.6 pricing should be validated against the official Azure OpenAI pricing page, as billing rates are determined by the meter associated with each model and deployment type. Microsoft pricing pages also note that pricing can vary by agreement, region, deployment type, and may be updated over time. [Azure Open...soft Azure | Azure.Microsoft.com], [azure.microsoft.com]

Additionally, the Azure OpenAI prompt caching documentation states that GPT-5.6 and later models introduce billable cache writes, whereas earlier models do not. Cache reads remain discounted, but cache writes are billed separately. The exact rates should align with the pricing meters published on the Azure pricing page. [learn.microsoft.com]

If your Cost Management export is showing 0.01 per 1M tokens for cached input, cache writes, input, and output across GPT-5.6 Sol, Terra, and Luna, we are not aware of any public documentation indicating that all token categories are intended to be billed at the same rate. Likewise, we cannot confirm from the published documentation whether this represents a promotional rate or a future re-rating scenario. [Azure Open...soft Azure | Azure.Microsoft.com], [learn.microsoft.com]

The best approach is to compare the exported meter IDs/SKUs with the pricing meters listed on the official pricing page and verify the billed usage in your subscription. If the rates continue to appear inconsistent with the published pricing information, we recommend opening an Azure Billing support request so the billing team can validate the associated meters and confirm whether the pricing is expected for those GPT-5.6 meters.

References

Azure OpenAI Service Pricing [Azure Open...soft Azure | Azure.Microsoft.com]

Prompt caching with Azure OpenAI [learn.microsoft.com]

Note: The public documentation confirms that GPT-5.6 introduces billable cache-write meters, but it does not state that input, output, cached input, and cache-write tokens should all be charged at the same rate. Therefore, a billing review would be required to determine whether the observed pricing is expected for your subscription.
The Max Coder 5 Reputation points

2026-07-17T16:16:30.4666667+00:00

Thank you. I have also checked the Azure Retail Prices API. The corresponding GPT‑5.5 meters are published with differentiated rates, but I cannot find the GPT‑5.6 Sol, Terra, or Luna meter IDs in the public catalog.

Since all GPT‑5.6 Global Standard token categories are currently rated at exactly $0.01 per 1M tokens, could this be escalated to the Azure OpenAI billing or product team for confirmation? I will also open a private billing support request containing the meter IDs and usage export.
Thanmayi Godithi 11,550 Reputation points Microsoft External Staff Moderator

2026-07-18T09:22:35.28+00:00

Hello The Max Coder ,

Thank you for checking the Azure Retail Prices API and sharing those findings.

At this time, there is no public documentation or pricing catalog entry that I can reference to confirm whether the GPT‑5.6 Sol, Terra, and Luna Global Standard meters are intentionally priced at $0.01 per 1M tokens across all token categories or whether this reflects a temporary pricing state.

Since the GPT‑5.6 meter IDs do not currently appear in the public retail pricing catalog and pricing/billing behavior cannot be validated through Microsoft Q&A, opening a private billing/support request with the meter IDs, usage export, subscription details, and timestamps is the appropriate next step. The support team can engage the Azure OpenAI billing and product teams as needed to verify the intended pricing and determine whether any future re-rating or billing adjustment is expected.
The Max Coder 5 Reputation points

2026-07-18T11:01:25.3466667+00:00
Thank you. One important detail is that this usage is occurring under a Microsoft Azure Sponsorship subscription funded by Microsoft for Startups credits.

Other models in the same subscription, including GPT‑5.5, are showing differentiated rates for cached input, input, and output, and those costs are being recorded normally against the sponsorship usage. GPT‑5.6 is the only model family where Sol, Terra, and Luna meters are uniformly rated at $0.01 per 1M tokens.

We have been monitoring the official Azure usage export specifically to manage our startup-credit consumption. Because the export consistently recorded a minimal GPT‑5.6 cost, that recorded cost materially influenced our testing and usage decisions.

Before continuing that usage, our primary concern is whether Microsoft could retrospectively apply corrected rates to GPT‑5.6 usage already recorded from July 9, 2026 onward.

I will open a private billing request with the meter IDs and export. Could you please confirm whether the billing/product team should specifically determine:

Whether the existing $0.01 rate is intentional.

Whether previously recorded usage can be retrospectively re-rated.

Whether any adjustment would be fully covered by startup credits because the usage occurred under an active sponsorship subscription.

Whether a corrected price could instead be applied prospectively once customers are notified.
Hendrik 30 Reputation points

2026-07-20T11:50:15.3833333+00:00

@The Max Coder Check your Cost Management! Azure shows now for the whole time cost and they was calculated complete wrong. They want to charge us a four-figure sum from one week usage just with a few request. We calculated the usage on our own and we spend about 100-150$ per day, cost management shows 7-8 times more per day. Also the ratio of cache write and input tokens is complete out of scope, as cache write is 25% from the input price, so input cost can be max 4x times of cache write cost.

You should contact immediately the support if something is wrong like for us
Arne 0 Reputation points

2026-07-21T12:36:31.53+00:00

I can confirm the same observation. The cost management reports cost significantly above my own calculation from the API returned token use for the gpt-5.6 models.

1 answer

Your answer

The Max Coder 5 Reputation points

2026-07-17T16:16:30.4666667+00:00

Thank you. I have also checked the Azure Retail Prices API. The corresponding GPT‑5.5 meters are published with differentiated rates, but I cannot find the GPT‑5.6 Sol, Terra, or Luna meter IDs in the public catalog.

Since all GPT‑5.6 Global Standard token categories are currently rated at exactly $0.01 per 1M tokens, could this be escalated to the Azure OpenAI billing or product team for confirmation? I will also open a private billing support request containing the meter IDs and usage export.
Thanmayi Godithi 11,550 Reputation points Microsoft External Staff Moderator

2026-07-18T09:22:35.28+00:00

Hello The Max Coder ,

Thank you for checking the Azure Retail Prices API and sharing those findings.

At this time, there is no public documentation or pricing catalog entry that I can reference to confirm whether the GPT‑5.6 Sol, Terra, and Luna Global Standard meters are intentionally priced at $0.01 per 1M tokens across all token categories or whether this reflects a temporary pricing state.

Since the GPT‑5.6 meter IDs do not currently appear in the public retail pricing catalog and pricing/billing behavior cannot be validated through Microsoft Q&A, opening a private billing/support request with the meter IDs, usage export, subscription details, and timestamps is the appropriate next step. The support team can engage the Azure OpenAI billing and product teams as needed to verify the intended pricing and determine whether any future re-rating or billing adjustment is expected.
The Max Coder 5 Reputation points

2026-07-18T11:01:25.3466667+00:00

Thank you. One important detail is that this usage is occurring under a Microsoft Azure Sponsorship subscription funded by Microsoft for Startups credits.

Other models in the same subscription, including GPT‑5.5, are showing differentiated rates for cached input, input, and output, and those costs are being recorded normally against the sponsorship usage. GPT‑5.6 is the only model family where Sol, Terra, and Luna meters are uniformly rated at $0.01 per 1M tokens.

We have been monitoring the official Azure usage export specifically to manage our startup-credit consumption. Because the export consistently recorded a minimal GPT‑5.6 cost, that recorded cost materially influenced our testing and usage decisions.

Before continuing that usage, our primary concern is whether Microsoft could retrospectively apply corrected rates to GPT‑5.6 usage already recorded from July 9, 2026 onward.

I will open a private billing request with the meter IDs and export. Could you please confirm whether the billing/product team should specifically determine:

Whether the existing $0.01 rate is intentional.

Whether previously recorded usage can be retrospectively re-rated.

Whether any adjustment would be fully covered by startup credits because the usage occurred under an active sponsorship subscription.

Whether a corrected price could instead be applied prospectively once customers are notified.
Hendrik 30 Reputation points

2026-07-20T11:50:15.3833333+00:00

@The Max Coder Check your Cost Management! Azure shows now for the whole time cost and they was calculated complete wrong. They want to charge us a four-figure sum from one week usage just with a few request. We calculated the usage on our own and we spend about 100-150$ per day, cost management shows 7-8 times more per day. Also the ratio of cache write and input tokens is complete out of scope, as cache write is 25% from the input price, so input cost can be max 4x times of cache write cost.

You should contact immediately the support if something is wrong like for us
Arne 0 Reputation points

2026-07-21T12:36:31.53+00:00

I can confirm the same observation. The cost management reports cost significantly above my own calculation from the API returned token use for the gpt-5.6 models.

Answer 1

Hello The Max Coder,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

Regarding your question:

Is GPT-5.6 Standard Global pricing currently 0.01 per 1M tokens?

The observed $0.01 per 1M tokens for GPT-5.6 Sol/Luna/Terra Global Standard is not publicly validated by official Microsoft pricing documentation. The official Azure OpenAI pricing page does not publish GPT-5.6 Sol/Luna/Terra at $0.01 across input, output, cached input, and cache write. Microsoft’s prompt caching documentation confirms that GPT-5.6 and later models bill cache writes separately from discounted cache reads, so a uniform $0.01 across all token categories is not supported by the public documentation. You will need to validate the exact ( MeterId, MeterName, SkuName, UnitOfMeasure, UnitPrice, PAYGPrice, and EffectivePrice from Cost Management export.) through Azure Retail Prices API. For example if the exact meter IDs are absent, the rate is not publicly verifiable and must not be used as confirmed production pricing. For invoice reconciliation only, use EffectivePrice × Quantity from Cost Management export.

Use the below resource links for more reading and steps:

I hope this is helpful. Please! Do not hesitate to let me know if you have any other questions, steps or clarifications.

Please do not close the thread by upvoting and accepting the answer if any part of it is helpful.

The Max Coder 5 Reputation points

2026-07-19T06:56:47.42+00:00

Thanks. This is a Microsoft for Startups Azure Sponsorship subscription, so the usual Cost Management export and Usage Details API aren’t available. I only have the CSV from the Azure Sponsorship Usage portal. It contains the resource GUID, service name, quantity, and cost, but not the unit, PAYG, or effective price.

I calculated the rate as Cost ÷ Quantity. All GPT‑5.6 sol, terra, and luna token categories work out to approximately $0.01 per 1M tokens. In the same export, GPT‑5.5 correctly works out to $0.50 for cached input, $5 for input, and $30 for output.

The GPT‑5.6 resource GUIDs also don’t appear in the Retail Prices API. Could someone from the billing or Azure OpenAI product team confirm whether this rate is intentional and whether usage already recorded at this rate might be re-rated later? If it is adjusted, would the startup credits active when the usage occurred cover that adjustment?
Sreetheja Adusumilli 965 Reputation points Microsoft External Staff Moderator

2026-07-20T04:53:59.6733333+00:00

Hello @The Max Coder ,

We sincerely apologize to hear that the issue has not yet been resolved. Your satisfaction is important to us, and we are committed to working with you toward a resolution. Please share your email ID, feasible time to connect in Private Message.