# Listed isn’t available. Call the model before you plan on it.

> Lab 031 · Editorial lab · Status: published · Author: Keith Townsend · November 19, 2026  
> Published by The CTO Advisor LLC (The Advisor Bench) · Layer2C Labs  
> Canonical: https://labs.layer2c.com/labs/reader-on-your-cloud · Structured: https://labs.layer2c.com/labs/reader-on-your-cloud.json

**Finding:** On my existing accounts, Claude Haiku 5.5 was callable with a stated limit above the Clinic’s rate only on Amazon Bedrock, at AWS’s default 10 million tokens a minute; Vertex AI returned zero-quota errors on two Google Cloud projects and denied an increase within 12 seconds, and Azure refused Microsoft’s own generally available Decision-1.1 at 0 of 0 quota.

**Question:** Every cloud lists the frontier models, and each announces its own as generally available (GA). I tried to call the Clinic’s readers through each cloud’s own model API, on accounts I already use. The catalog and the GA label said nothing about whether my account could call the model.

**Load:** The Clinic is my scheduling bench: code runs a simulated clinic’s booking conversations, and when it can’t parse a patient’s reply it asks a reader model a closed question. A cell is one run of 48,000 conversations. The rate bar comes from the three Claude Haiku 5.5 cells in arm E, where the reader’s yes counts as consent: about 1,300 requests and 400,000 input tokens a minute.

**Raw lab detail:** https://labs.layer2c.com/downloads/reader-on-your-cloud-lab-detail.md
**Vendors:** Microsoft (cloud), AWS (cloud), Google Cloud (cloud), Anthropic (model), OpenAI (model), Google (model)
**Layers:** Layer 2C (Reasoning)
**Themes:** managed-abstraction-cost, vendor-claim-scrutiny

## Verdict

**Scope:** What this covers: accounts I already use on each cloud, US regions, November 18 and 19, 2026. On Azure, a pay-as-you-go subscription I’ve had for years that had never used Foundry, Azure’s model service. On AWS, my lab account. On Google Cloud, two projects with production and test workloads. One call per reader on each route, a quota read against the Clinic’s rate, and one quota request where a cloud offered it. That shows whether a model is callable and what limit the cloud states, not sustained throughput. API access to hosted models only. It doesn’t cover Anthropic’s own API, which is how the Clinic has always reached Claude, or an Enterprise Agreement or committed-spend account. The open question: what does an enterprise agreement start on?

- **DON'T:** Don’t read a GA announcement or a catalog listing as access. Microsoft-Decision-1.1 went GA on November 17 and was listed in every region I read. On my subscription its quota was 0 of 0, and Azure refused the deployment both times I tried.
- **DO:** Before you plan on a model through your cloud, make one real call and read its quota against your workload’s rate. Here that was a handful of reads and calls per cloud, a few hundred tokens in all, and it settled what the catalogs couldn’t.
- **DON'T:** Don’t assume your cloud will let you call a frontier model it resells. Claude Haiku 5.5 answered on Bedrock on the first call. Vertex AI returned a zero-quota error on both projects I tried and denied the increase request in seconds. Azure stopped the deployment for company details, with zero quota behind that.
- **DO:** If the reader has to move between clouds, test every cloud at your rate, not just for a reply. gpt-oss-120b was the only reader all three accepted, and only AWS and Azure stated a limit above the bar. Vertex AI’s limit for it didn’t read.

**Independence and disclosure:** Self-funded editorial. Google Cloud is a client of The CTO Advisor LLC. Microsoft and AWS are vendors under test. Anthropic makes Claude, the control reader. An Anthropic model ran the scripted checks under my direction and drafted this record; the evidence is each cloud’s own API responses, kept raw. None of Microsoft, AWS, Google or Anthropic reviewed this page.

## Measured

| Measurement | Result | Note |
| --- | --- | --- |
| The bar | About 1,300 requests and 400,000 input tokens a minute | From the three Claude Haiku 5.5 arm E cells at 16 parallel sessions: 1,136 to 1,301 requests and 329,000 to 382,000 input tokens a minute, about 310 tokens a call, 25 to 29 million tokens a cell, 70 to 81 minutes each. |
| Claude Haiku 5.5 on Amazon Bedrock | Answered; stated limit 10 million tokens a minute | Equal to AWS’s default, as are Sonnet 5.5 (6 million) and Opus 5.5 (30 million). Bedrock states no requests-per-minute quota for Haiku 5.5, only tokens. One call, first try. |
| Claude Haiku 5.5 on Vertex AI | Zero-quota error on both projects | Quota exceeded for anthropic-claude-haiku at the global endpoint, on the test project and then on the production project. The quota read beforehand returned no value for any Anthropic model, which turned out to mean zero. The model isn’t offered in us-east5 (404). |
| The Vertex AI quota request | Denied within 12 seconds | Filed through the Cloud Quotas API on the test project for 1,500 requests, 500,000 input tokens and 100,000 output tokens a minute, global, at 10:55 AM Central on November 19. All three limits read eligible for an increase before filing. No written justification was sent. Denials are stamped 4 to 12 seconds after filing. |
| Claude Haiku 5.5 on Azure AI Foundry | Refused at deployment | Anthropic deployments require company details (industry, organization name, country) before the quota check. I didn’t submit them. The quota read shows zero for all 13 Anthropic models. |
| Azure Foundry’s starting tier | Free Tier on both subscriptions | Foundry sorts subscriptions into quota tiers, Free and 1 through 6. Both were set to Free when the Foundry provider was first registered, November 18. In eastus2, 56 of 92 GA models with a quota line have zero quota, including all 13 Anthropic models and 41 of 50 OpenAI models. Free Tier can’t request an increase, and there’s no API to file one. |
| Microsoft-Decision-1.1 | Refused at deployment, 0 of 0 quota | GA November 17. Refused at the smallest capacity twice: a Data Zone deployment (served within one geography) on November 18, and a Global deployment on November 19. Available capacity read zero in all 55 region and deployment-type entries. |
| The cloud’s own model on AWS and Google Cloud | Both answered | Nova 2 Lite on Bedrock: 2,000 requests and 8 million tokens a minute. Gemini 3.8 Flash on Vertex AI: 50 million input tokens a minute and 5 billion a day; its requests-per-minute bucket returned no value, so its request limit isn’t known. |
| gpt-oss-120b | Answered on all three | Bedrock: 100 million tokens and 10,000 requests a minute. Azure: 5 million tokens a minute of quota, though the call ran on a deployment at the smallest capacity, 1,000 tokens a minute, deleted afterward. Vertex AI: answered at two endpoints; its quota read returned no value, so its limit isn’t known. |
| Other open-weight models on Azure | 20,000 tokens a minute | Llama, Phi, Mistral and DeepSeek, each about 20 times under the bar. A run under that limit only finishes if the runner is paced to it: a 75-minute cell would take about a day. At 16 parallel sessions most calls would be refused. |
| Gemma 4 26B and 31B | Not served GA per token anywhere | Not on Bedrock in three US regions. Public preview only on Vertex AI. On Azure, listed only for Microsoft-run compute on dedicated GPUs, not per token. |

## Authority placement (DAPM)

**Reading:** Decision authority: who makes the runtime decision at this layer, and can the enterprise see and override it?

| Layer | Azure AI Foundry | Amazon Bedrock | Google Cloud Vertex AI | Basis |
| --- | --- | --- | --- | --- |
| Layer 2C (Reasoning) | Ceded | Ceded | Ceded | Who decides whether the reader runs at my rate, not who decides inside it. On every route the cloud does, through its quota and entitlement policy, and the same model behind the same API got different answers. Bedrock granted Claude by default; Vertex AI set it to zero and denied the increase; Azure set Microsoft’s own model to zero with no way to request more. A granted limit is still the cloud’s decision. It’s Ceded where the call worked too, which is why the check belongs in the plan. |

## What this does not establish

- Throughput. One call per reader shows a model is callable and a quota read shows the stated limit; neither shows the cloud sustains the bar.
- Anything about an Enterprise Agreement, committed-spend or partner-managed account.
- What Azure does with Claude once the company details are submitted.
- That Claude is gated outside the routes I tried, or that partner models other than Claude are.
- Anything about running models on rented GPUs, or about the models’ quality. Every call was a one-word reply.

## Open questions

- **What does an Enterprise Agreement or committed-spend account start on?** Open. I haven’t had one to test. On Azure it decides whether Free Tier is the brownfield default or a pay-as-you-go artifact.
- **Would a justified or smaller request on Vertex AI be granted, and how fast?** Open. The denial answered one plain request at the full rate on one project. A written case, a smaller first step, or a project with Vertex AI spend could get a different answer.

## Objection handled

**These are small accounts. Wouldn’t an enterprise on an Enterprise Agreement get real quota?**

It might, and I haven’t tested it. Microsoft’s documentation says Foundry’s starting tier weighs the customer’s relationship with Microsoft, and an agreement is exactly that. My subscriptions are pay-as-you-go, years old, and had never used Foundry, and both started on Free Tier when I first turned the service on. That shows years of other Azure spend didn’t lift the starting tier. It doesn’t show what an agreement starts on.

AWS doesn’t settle it either. Its Claude defaults are already 25 times the Clinic’s bar, so a new account and an old one look the same there, and my account has 23 limits set below AWS’s defaults on other lines. Bedrock tells you nothing about a cloud whose starting limit is zero.

So I can’t say an agreement gets nothing. I can say the check doesn’t depend on it. Whatever your agreement says, one call and one quota read tell you what your account can do, for almost nothing. Run them before the architecture review, not after.

## Sources

- [Microsoft: Microsoft-Decision-1.1 general availability announcement](https://commandline.microsoft.com/microsoft-decision-model-ga/)
- [Microsoft Learn: Foundry Models quotas and limits](https://learn.microsoft.com/en-us/azure/foundry/foundry-models/quotas-limits)
- [AWS: Amazon Bedrock quotas](https://docs.aws.amazon.com/bedrock/latest/userguide/quotas.html)
- [Google Cloud: generative AI on Vertex AI quotas](https://cloud.google.com/vertex-ai/docs/generative-ai/quotas-genai)

## Summary

The Clinic reaches Claude through Anthropic’s own API. Plenty of enterprises can’t. They buy through the cloud they already have an agreement with. So I asked whether the Clinic’s reader could run through each cloud’s own model service, on accounts I already use, at the Clinic’s rate.

Only one route cleared both tests, an accepted call and a stated limit above the bar: Claude Haiku 5.5 on Amazon Bedrock. Google Cloud and Azure both blocked Claude on these accounts, at different points. Each cloud’s own model was callable on AWS and Google Cloud and blocked on Azure. gpt-oss-120b was the one reader all three accepted.

## Detail

This started as a deployment failure. I wanted Microsoft-Decision-1.1, announced GA the day before, as a reader in the Clinic, and Azure refused it at 0 of 0 quota. So I built a ladder I could run on all three clouds: the catalog, the entitlement (terms, forms and restrictions on this account), the quota against the Clinic’s rate, and then one real call. The first three rungs passing doesn’t mean the fourth will.

The bar matters because a limit under it doesn’t always stop a run. It stretches it, if you pace the runner, and a daily cap under about 30 million tokens stops a cell outright.

Azure was the strictest. Free Tier zeroes out most of the frontier catalog, Microsoft’s own Decision-1.1 included, and it can’t request more. gpt-oss-120b, with 5 million tokens a minute of quota, was the one reader there with a stated limit above the bar.

AWS accepted every reader I tried. Claude Haiku 5.5, Nova 2 Lite and gpt-oss-120b all answered on the first call, as did Gemma 3 27B, which is out of the comparison because Gemma 4 isn’t offered. AWS does tighten some limits per account: my cross-model daily token cap is 310 million against a 700 million default. That still covers about ten cells a day.

Google Cloud split. Gemini 3.8 Flash and gpt-oss-120b answered. Claude hit zero quota on both projects I tried. The quota API said the Claude limits were eligible for an increase, which Azure’s Free Tier isn’t, so I filed one on the test project at the Clinic’s rate, with no justification attached. It came back denied in seconds. Google gave me a form and an automated no; Azure gave me no form. They stopped me at different points, and neither got me Claude.

## Method and disclosure

Self-funded editorial. Reads came from each cloud’s own APIs and command-line tools: the model catalog, per-account restrictions and quota limits, and on Google Cloud the Cloud Quotas API for the increase request. Each reader got one call per route, “Reply with the single word: ready,” with 16 output tokens at most, plus a retry where the first failure was mine: Claude on Vertex AI was first sent to a region that doesn’t offer it, and the first Azure gpt-oss-120b call lacked a permission on my side. The reasoning models (gpt-oss-120b, Gemini 3.8 Flash) spent their 16 tokens thinking and returned no answer text; I count an accepted, billed request as callable. On Azure, deploying a model comes before a call, so each reader was deployed at the smallest capacity, called if it deployed, and deleted. Google Cloud quotas were read on the production project; the calls ran on the test project and Claude again on production; the quota request ran on the test project. Reads and the first Decision-1.1 deployment attempt ran November 18, the calls and the quota request November 19, 2026, all in US regions. The raw responses go in the lab detail download.

---
*Layer2C Labs · The CTO Advisor LLC (The Advisor Bench) · labs.layer2c.com*
