OpenAI APIvs
AWS SageMaker
AWS SageMaker: Build, train, and deploy machine learning models at scale
OpenAI API
Hosted API for OpenAI's language, embedding, image and audio models, billed per token
As of 30 August 2026, OpenAI API starts at $0.15/per-million-tokens. You send text, images or audio to an endpoint and are billed per token, with no weights to download and no on-premises option. Softwr lists it under Machine Learning. OpenAI API is made by OpenAI, launched in 2015, available on API.
Overview
What OpenAI API does
The OpenAI API is a hosted service reached over HTTP. Requests are billed per token of input and output at rates that differ by model, with a cheaper batch path for work that can tolerate delay and a discount on cached input prefixes. The surface covers text generation and reasoning models, embeddings, image generation, speech to text and text to speech, alongside server-side facilities such as function calling, structured output schemas and file search. Model weights are not distributed and there is no on-premises deployment; the nearest equivalent is Microsoft's Azure OpenAI Service, which serves the same model family under a different contract and a different data processing agreement. The distinguishing property is that you inherit somebody else's product decisions. Models are superseded, older snapshots are retired on a published schedule, and behaviour changes between versions in ways that no version string in your code will warn you about. What you get in exchange is capability that no organisation can currently reproduce on its own hardware at the same quality for the same money. That is the trade every buyer here is actually making, and it is worth being explicit about it: capability now, in return for a dependency whose roadmap, pricing and availability sit entirely outside your control. The buyers are essentially every product team adding language features, because the alternative is a research programme most organisations cannot staff. The trade-off is that cost scales with usage rather than with seats, so a feature that succeeds raises the bill in proportion to its success and the unit economics have to be worked out before launch rather than after. Fine-tuned models are attached to a base model that will eventually be retired, prompts are tuned against behaviour that will change, and an evaluation suite is not an optional nicety but the only thing that makes a model migration survivable.
What people use it for
- Adding summarisation, drafting or classification to an existing product where building a model would take longer than the product's whole roadmap
- Retrieval-augmented question answering over internal documents, using the embedding and generation models together
- Extracting structured records from unstructured text, where schema-constrained output removes most of the parsing problem
- Prototyping a language feature quickly to find out whether it is worth the cost of a self-hosted alternative later
The honest half
Where it falls short
Concrete and checkable, so you can decide whether any of them matter to you. This is the half of a review a vendor will not write about OpenAI API.
- Cost scales with tokens rather than with seats, so a successful feature's bill grows with its adoption, and an interface that lets users paste long documents has no natural ceiling on spend unless you build one yourself.
- Models are deprecated on the vendor's timetable, and a fine-tuned model built on a retired base goes with it, so the tuning work and the data curation behind it must be redone rather than migrated.
- Behaviour shifts between model versions in ways no test catches unless you wrote one, so prompts tuned over months against a particular snapshot can regress quietly on migration, which makes an evaluation suite a prerequisite rather than an improvement.
- It cannot run inside your own network, so data residency requirements, air-gapped environments and contracts forbidding third-party processing rule it out regardless of the provider's own security posture.
- You inherit its availability and its rate limits, so a provider incident is an outage in your product and a traffic spike can be throttled at precisely the moment the feature is proving itself.
Cross-shopped
What people choose instead of OpenAI API
Each pairing was judged by two reviewers asking whether a buyer would genuinely weigh the two against each other. The ones that failed were deleted rather than published.
- OpenAI APIvsGoogle Vertex AI
Google Vertex AI: Unified ML platform to build, deploy, and scale AI models
- OpenAI APIvsAzure Machine Learning
Azure Machine Learning: Enterprise-grade machine learning service
- OpenAI APIvsDataRobot
DataRobot: Enterprise AI platform for automated machine learning
Pricing
What OpenAI API costs
Taken from the vendor's own pricing page. Prices move, so check before you buy.
GPT-4o mini
$0.15 /per-million-input-tokens
- Fast
- Affordable
GPT-4o
$5 /per-million-input-tokens
- Multimodal
- 128K context
Capabilities
Features
Text and reasoning models
Generation, instruction following and multi-step reasoning over text and images
Embeddings
Vector representations for search, clustering and retrieval pipelines
Speech and audio
Transcription, translation and text to speech through the same account and billing
Image generation
Creation and editing of images from text prompts
Function calling
The model returns structured calls against tool schemas you define, for wiring into your own code
Structured outputs
Responses constrained to a JSON schema so downstream parsing does not depend on prompt discipline
Batch processing
A cheaper asynchronous path for work that does not need an immediate response
Prompt caching
Reduced cost on repeated input prefixes, which matters for long system prompts
Fine-tuning
Supervised tuning of selected base models on your own examples, hosted on the platform
Usage tiers and rate limits
Throughput allowances that rise with account history and spend
Answered, with sources
Questions people ask
Each answer names the page it came from, so you can check it rather than take our word for it.
Is my data used to train the models?
API inputs and outputs are not used for training by default, which differs from the consumer product. Retention periods and enterprise terms change, so read the current data usage policy rather than trusting a summary.
Can I run these models on my own hardware?
No. The weights are not distributed. If self-hosting is a requirement, you are looking at open-weight models instead, with the operational and quality trade-offs that implies.
How is it priced?
Per token, with input and output priced differently and each model priced differently. Batch processing and cached input prefixes reduce it. The practical consequence is that your bill is a function of prompt design, not just of request count.
What is the difference from Azure OpenAI Service?
The same model family delivered by Microsoft under an Azure contract, with Azure identity, networking and regional controls, and a different release cadence for new models. Enterprises with an Azure agreement often choose it for procurement and data residency reasons rather than technical ones.
How do I keep the cost under control?
Cap input length, cache repeated prefixes, route easy requests to smaller models, use the batch path where latency does not matter, and set per-user limits before launch rather than after the first surprising invoice.
Behind it
Who makes OpenAI API
- Company
- OpenAI
- Based in
- San Francisco, California, USA
- Founders
- Sam Altman, Elon Musk, Ilya Sutskever, Greg Brockman, Wojciech Zaremba, John Schulman, Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, Pamela Vagata
Keep looking
Where to go from OpenAI API
- OpenAI API alternativesEvery product people weigh against it
- OpenAI API pricingPlans, floors and what the tiers exclude
- OpenAI API badgesWhich rules it meets, and the figure behind each
- All Machine Learning softwareThe full category
- Best Machine Learning softwareThe shortlist, and how it was cut
Best Machine Learning software for
Compare OpenAI API with
- Cohere
- AWS SageMaker
- Google Vertex AI
- Azure Machine Learning
- DataRobot
- Fal AI
- BentoML
- Snowflake
- Hugging Face
- Python
- Ollama
- Neptune.ai
- Weka
- ClearML
- BigQuery ML
- Semantic Kernel
- BentoML
- Semantic Kernel
Other Machine Learning software
Build, train, and deploy machine learning models at scale
Free, then $0.04/hour9 researched notesMicrosoft's managed platform for training, tracking and deploying models on Azure
Free plan13 researched notesOpen source Python framework that packages models into deployable inference services
Free plan14 researched notesOpen-source MLOps platform for experiment tracking and orchestration
Free plan8 researched notesOpen source Python framework that packages models into deployable inference services
Free plan14 researched notes
Softwr does not host reviews and shows no star rating for OpenAI API, because a rating we did not collect is not ours to publish. What is here is the pricing and platform detail from the vendor’s own pages, limitations we could state concretely, and alternatives a reviewer confirmed people weigh against it. Tell us if any of it is wrong.
More on OpenAI API
OpenAI API alternatives
What people switch to, and what they give up
OpenAI API pricing
Every tier, and where the cost actually lands
Compare OpenAI API
Put it head to head with anything we hold
OpenAI API badge
Its rating, and an embed for your own site
Best Machine Learning software alternatives
Open-source MLOps platform for experiment tracking and orchestration
Open-source self-hosted, with paid hosted and enterprise tiers
LLM engineering platform for testing and evaluating AI agents in production
Tiered subscription with usage-based overage charges
Open-source AI orchestration framework for LLM applications
Open-source with optional paid enterprise support
The world's most popular data science platform