Where Each Model Is Actually Available
Where OpenAI's models run, and what changes off the direct API
| OpenAI direct API | The baseline every figure on this site is sourced from |
|---|---|
| Amazon Bedrock — pricing | Billed through AWS; matches OpenAI direct pricing in commercial regions. Region-specific use is priced like OpenAI's Regional processing. |
| Bedrock — GPT-6 Sol, GPT-6 Luna | Bedrock Runtime (US and global inference profiles) and Mantle in us-east-1 |
| Bedrock — GPT-6 Astra | Bedrock Runtime and Mantle in us-west-2 |
| Bedrock context window — GPT-5.4, GPT-5.5 | 1,000,000 tokens |
| Bedrock context window — GPT-5.6 Sol, Terra, Luna, GPT-6 Astra | 1,050,000 tokens |
| Not available on Bedrock | Asynchronous tool calling (async: true); WebSocket connections; Mid-turn steering; Reasoning updates (configuration_update input items); Pro mode; Programmatic Tool Calling; Multi-agent; Hosted file search; Shell tool; Image generation tool; Remote MCP servers |
| Billed through the provider | Amazon Bedrock, Microsoft Azure |
| Batch API — platforms named as lacking it | None named on any OpenAI page checked (see the note below) |
Models listed by OpenAI that no calculator here offers
| GPT-5.6 Cyber | Requires separate approval and provisioning through OpenAI's Daybreak program (Trusted Access for Cyber). Not offered in any calculator on this site. |
|---|---|
| GPT-5.5 Cyber | Cyber models require Daybreak approval. Not offered in any calculator on this site. |
| GPT-5.4 Cyber | Deprecated on 2026-09-11 and removed from the API on 2026-10-01. OpenAI's stated replacement is "the most capable cyber model available to you". |
"Available" isn't one fact about a model — it's available somewhere, at some price, with some set of features, and the October re-check changed what OpenAI says about all three.
Bedrock pricing now matches — in commercial regions
This page used to say Amazon Bedrock's pricing differs from OpenAI's direct rates. OpenAI's own Bedrock guide now says the opposite: Bedrock pricing in commercial regions matches OpenAI direct pricing for equivalent services, and a region-specific Bedrock service is priced like Regional processing on the OpenAI API — which carries its own uplift. Billing still runs through AWS, and AWS's own pricing page and model cards remain the authority on which service tiers Bedrock offers.
What still differs on Bedrock
Context windows, for two models: on Bedrock, GPT-5.4 and GPT-5.5 get a smaller window than they have on OpenAI direct, while the GPT-5.6 models and GPT-6 Astra keep the full window. Model placement: GPT-6 Sol and Luna run through Bedrock Runtime and the Mantle endpoint in one US region, GPT-6 Astra in another. And features: asynchronous tool calling, WebSocket connections, mid-turn steering, reasoning updates, pro mode, Programmatic Tool Calling, multi-agent, the shell and image-generation tools, hosted file search and remote MCP servers are all unavailable there, and hosted web search works only through the Mantle endpoint. The newest models, GPT-6.1 Sol and GPT-6.1 Terra, aren't named in OpenAI's Bedrock guide at all.
What this site no longer claims
It no longer lists Batch as unavailable on Bedrock, Google Vertex AI and Microsoft Foundry. No OpenAI page checked in October names any platform as lacking Batch, and Vertex and Foundry don't appear on OpenAI's pages at all; the pricing page names Bedrock and Microsoft Azure as platforms that bill through their own services. If Batch matters to your cost model on a cloud platform, confirm it in that platform's documentation before designing around it.
The models no calculator offers
The second table lists models OpenAI publishes that this site won't price: the approval-gated Daybreak cyber models, and models removed from the API.
Verified 2026-10-28 against https://developers.openai.com/api/docs/guides/amazon-bedrock.
Could not confirm: Google Vertex AI and Microsoft Foundry: no OpenAI page checked on 2026-10-28 mentions either. Which service tiers (Batch, Flex, Fast) Bedrock offers: OpenAI's guide defers that to AWS's own pricing page and model cards. GPT-6.1 Sol and GPT-6.1 Terra on Bedrock: the guide names neither.
Checked: https://developers.openai.com/api/docs/guides/amazon-bedrock · https://developers.openai.com/api/docs/pricing · https://developers.openai.com/api/docs/guides/batch
Verified 2026-10-28 against https://developers.openai.com/api/docs/guides/batch.
Could not confirm: Which cloud platforms lack Batch. Until 2026-10-01 this site listed Amazon Bedrock, Google Vertex AI and Microsoft Foundry; none of the OpenAI pages checked on 2026-10-01 names any platform as lacking Batch, and the Bedrock guide defers service tiers to AWS's own pricing page. So no platform is listed — check your provider's docs.
Checked: https://developers.openai.com/api/docs/guides/batch · https://developers.openai.com/api/docs/pricing · https://developers.openai.com/api/docs/guides/amazon-bedrock