Niches · Model APIs & inference

Call frontier models inside a cloud account and get capacity

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

0companies, full fit
4companies, fit with a limit
42companies carry the tag
6share of the category's work
3first-hand accounts

What counts as done

The first call succeeds within a day of enabling the model, and a 429 says whether it was quota or shared capacity.

Measured, for our own tests: a steady 10 requests a minute runs for an hour without a capacity 429. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: use Claude through Vertex or Bedrock under our existing bill, why is a low-volume Gemini call getting 429 on the global endpoint, our Claude quota on Vertex is 0

Does the job, with a limit 4

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
Amazon Bedrock A
Amazon Web Services, Inc.
Quotas are per model per Region and shared across APIs, 429 ThrottlingException is not documented as quota or capacity, increase requests may be denied without existing usage, and Provisioned Throughput pricing per model unit goes through an account manager. source
Claude API A
Anthropic, PBC
Claude Platform on AWS bills through AWS Marketplace on the Start tier with the same rate limits; tier increases go through an account representative, and we read only the rate-limit page's summary of it. source
Azure OpenAI in Foundry Models A
Microsoft Corporation
Quota tiers rise automatically with spend and Tier 0 cannot request increases; PTU quota does not guarantee regional capacity and a deployment fails if capacity is short; a runtime 429 is documented as TPM only, not as quota versus capacity. source
Vertex AI (Gemini Enterprise Agent Platform) AA
Google LLC
429 text distinguishes 'Resource exhausted, please try again later' (shared capacity) from 'Exceeded the Provisioned Throughput'; Claude quotas are per-model QPM buckets of 1,000 to 1,500, yet acc-27 reports quotas granted at 0 pending review and acc-26 low-volume 429s. source

Couldn't check 2

The docs were closed to us, or silent. These carry no mark either way.

ListingWhy
StepFun API E
STEPFUN TECHNOLOGY (HK) LIMITED
The host platform.stepfun.ai closed the connection on every attempt (curl exit 56, no HTTP status) from this session, so no page was read.
Tencent Cloud TokenHub C
Tencent Cloud Computing (Beijing) Co., Ltd. (腾讯云计算(北京)有限责任公司)
The host cloud.tencent.com closed the connection on every attempt (curl exit 56, no HTTP status) from this session, so no page was read.

Doesn't do this job 35

Listings in the category that do something else, with the reason.

ListingWhy
Antseed C
Antseed Foundation
Peer-to-peer network paid in USDC, not a cloud account route.
BlockRun.AI A
BlockRun, Inc.
Pay-per-call gateway over x402, not a cloud account route.
Cerebras Inference B
Cerebras Systems Inc.
Not a cloud account route; dedicated inference is by agreement.
Cloudflare AI Gateway B
Cloudflare, Inc.
A proxy over other providers, not a cloud account route.
Cloudflare Workers AI BB
Cloudflare, Inc.
Not a cloud account route in the map's sense; capacity 429s are code 3040 and there is no reserved capacity to buy.
Cohere Chat API (Command models) B
Cohere Inc.
Not a cloud account route; private deployments go through sales.
DeepInfra B
Deep Infra Inc.
Not a cloud account route.
DeepSeek API D
Hangzhou DeepSeek Artificial Intelligence Co., Ltd.
Not a cloud account route.
Eden AI C
Eden AI SAS
Not a cloud account route.
Glimpse F
Elorian AI, Inc.
Not a cloud account route.
Fireworks AI Inference BB
Fireworks.ai, Inc.
Not a cloud account route; reserved throughput is a separate product.
Gemini Developer API B
Google LLC
The Developer API is not the cloud account route; that is Vertex AI.
GroqCloud BB
Groq LLC
Not a cloud account route.
Hugging Face Inference Providers C
Hugging Face, Inc.
Not a cloud account route.
Jais 3 C
Mohamed bin Zayed University of Artificial Intelligence (Institute of Foundation Models), with Inception42 (a G42 company)
Not a cloud account route.
LiteLLM B
Berrie AI Incorporated
Self-hosted gateway, not a cloud account route.
Meta Model API D
Meta Platforms, Inc.
Not a cloud account route.
MiniMax API C
Nanonoble Pte. Ltd.
Not a cloud-account offering; MiniMax sells direct API keys and M Plan subscriptions only. source
Mistral AI API B
Mistral AI (RCS Paris 952 418 325)
Mistral's own API; cloud marketplace routes are outside these docs. source
Muse Spark Open C
Meta Platforms, Inc.
Downloadable weights, not a cloud-account service. source
Nemotron 3.5 Super VL C
NVIDIA Corporation
NVIDIA's catalogue is a trial, not a cloud account; production needs a partner host or an AI Enterprise licence at $4,500 a GPU a year. source
Novita AI C
Novita AI
A direct hosted API, not a cloud account. source
OpenAI API AA
OpenAI OpCo, LLC
The direct API; Azure routes are outside these docs. source
OpenRouter B
OpenRouter, Inc.
A router over providers, not a cloud account. source
Prism Inference C
Prism Technologies Inc
A direct prepaid API, not a cloud account. source
Router.com (Ramp Router) C
Ramp Business Corporation
A gateway with its own credit, not a cloud account. source
Beam 1.0 C
Reflection AI
A direct API, not a cloud account. source
RouterPlus D
EVO HQ Inc
A gateway with prepaid credit, not a cloud account. source
SambaCloud B
SambaNova Systems, Inc.
A direct hosted API, not a cloud account. source
SiliconFlow E
SiliconFlow Labs Pte. Ltd.
A direct hosted API, not a cloud account. source
Simplismart F
Verute Technologies Private Limited
Shared endpoints and dedicated GPUs on Simplismart's platform, not a cloud account. source
SWE-2 Flash F
Cognition AI, Inc.
Not a cloud-account offering. source
Together AI Inference B
Together Computer, Inc.
A direct hosted API, not a cloud account. source
Vercel AI Gateway B
Vercel Inc.
A gateway with its own credits, not a cloud account. source
xAI API (Grok) BB
SpaceXSI (xAI)
A direct API; the Azure Foundry route is Microsoft's offering. source

6 listings in the category were added after this reading and aren't marked yet.

What stops agents finishing

  • quota granted at 0 pending a manual access review with no stated timeline
  • shared-capacity 429s that backoff over 30 minutes does not clear, with provisioned throughput priced around 2,000 EUR a month minimum
  • account, IAM and region set-up that an agent cannot do alone

Around it

After: inference.tool-loop, inference.failover-routing.

Where this comes from

Weight 6 of 100 in Model APIs & inference: 3 of 28 accounts; two forum threads with no staff reply and one production team using cloud routes only as outage fallback

The same as data: /api/v1/use-cases/inference.cloud-quota.json, or the MCP tool get_use_case with id inference.cloud-quota. Other jobs in this category: /api/v1/niches?category=inference.

Draft, read 2026-11-12 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.