whichdevtool?
Menu
StacksDirectory Pricing changes Glossary
VPS and cloud servers App hosting and PaaS Managed databases Transactional email APIs Email marketing and newsletters Domain registrars CDN, DNS and edge security Monitoring, uptime and error tracking Auth and identity Frontend and full-stack frameworks Backend frameworks UI component libraries AI model APIs File and object storage Payments and billing Background jobs, queues and workflows ORMs and database toolkits

AI model API (model maker) · by OpenAI

OpenAI API

Pay-per-token API for OpenAI's GPT-5.x, image, realtime voice and embedding models, with Batch, Flex and Priority processing modes priced separately.

Visit openai.com ↗ Alternatives Paid
OpenAI API homepage
§1

Overview

The OpenAI API gives programmatic access to OpenAI's own models: the GPT-5.x text and reasoning family (standard, mini, nano and pro variants), gpt-image-2 for images, GPT-Realtime-2 for voice, streaming transcription and translation, embeddings and a free moderation model. Everything is billed per token or per unit of media, with no monthly subscription for API use.

Price depends on the processing mode as much as the model. Standard is the default; Batch and Flex cost half as much in exchange for asynchronous or slower, occasionally unavailable processing; Priority costs roughly twice as much for steadier latency. Cached input tokens are billed at about 10% of the input price, which matters for chat apps and agents that resend long prompts.

Before paying, note that spend is capped by usage tier: new accounts are limited to $100 a month until they have paid more, and the limit rises automatically as cumulative payments grow. Built-in tools such as web search and code containers are billed on top of tokens. Enterprise features such as data residency, Scale Tier and reserved capacity go through sales.

§2

Pricing and plans

Plan Price What you get
GPT-5.5 $5 in / $30 out per 1M Cached input $0.50; prices for context under 272K tokens
GPT-5.4 $2.50 in / $15 out per 1M Cached input $0.25
GPT-5.4 mini $0.75 in / $4.50 out per 1M Cached input $0.075
GPT-5.4 nano $0.20 in / $1.25 out per 1M Cached input $0.02
gpt-5.6-luna $1 in / $6 out per 1M Listed on the docs pricing page with cached input $0.10
Batch / Flex 50% off Batch runs asynchronously within 24 hours; Flex trades speed and availability for price
Priority About 2x standard e.g. GPT-5.4 at $5 in / $30 out per 1M

Prices are per 1M tokens in USD, read 2026-09-29 on openai.com/api/pricing and the docs pricing page; they change often. Data residency and regional processing add 10%. Hosted tools cost extra: web search $10 per 1,000 calls, containers from $0.03 per 1 GB per 20-minute session, file search storage $0.10 per GB per day after 1 GB. Usage tiers cap monthly spend: $100/month at the entry tiers, rising to $200,000/month at Tier 5 ($1,000 paid). OpenAI's rate-limit page lists a 'Free' tier for allowed geographies, but no free token allowance is published on the pricing pages, so none is recorded here. OpenAI is winding down its fine-tuning platform; it is closed to new users.

Source: openai.com/api/pricing/ · prices as published, excluding tax.

§3

What it does

Batch API
Asynchronous jobs completed within 24 hours at 50% of standard input and output prices.
Flex and Priority processing
Flex halves prices for non-urgent work with slower responses; Priority costs more for steadier performance on pay-as-you-go.
Prompt caching
Cached input is billed at a tenth of the base input price on current GPT-5.x models.
Hosted tools
Web search, file search and code containers run on OpenAI's side and are billed per call, per GB-day or per session.
Realtime and audio
GPT-Realtime-2 handles live voice; GPT-Realtime-Whisper transcribes at $0.017 per minute.
Usage tiers
Tiers 1 to 5 raise rate limits and the monthly spend ceiling as cumulative payments grow from $5 to $1,000.
§4

Strengths and limitations

Strengths

  • Three price levels per model (Batch/Flex, Standard, Priority) let you trade latency for cost without switching vendor.
  • Small models are cheap: gpt-5.4-nano costs $0.20 in and $1.25 out per 1M tokens.
  • Text, image, voice, transcription and embeddings under one key and one bill.
  • Published monthly usage limits per tier make the maximum exposure of a new account predictable.

Limitations

  • No free token allowance is published; you pay from the first request.
  • New accounts are capped at $100 of usage per month until cumulative payments move them up a tier.
  • Only OpenAI's own models are served; open-weight models from other labs are not available here.
  • The fine-tuning platform is being wound down and is closed to new users.
  • Data residency and regional processing add 10% to token prices.
§5

Who it suits

Good for

  • Products built around OpenAI's GPT-5.x models
  • Voice agents and realtime audio apps
  • Bulk classification or extraction jobs that can use the 50% Batch price

Look elsewhere if

  • Hobby projects that need a free allowance to start
  • Teams that want open-weight models or several vendors behind one key (use a router or an inference provider)
§6

Pricing history

  • OpenAI containers billed per 20-minute session

    From 2026-03-31, hosted containers (Code Interpreter, hosted shell) are billed per 20-minute session per container, from $0.03 for 1 GB, instead of per container. Source ↗

  • Assistants API deprecated

    Announced 2025-08-20: the Assistants API shut down on 2026-08-26; OpenAI points users to the Responses and Conversations APIs. Source ↗

  • GPT-4.5 preview deprecated

    Announced 2025-04-14: gpt-4.5-preview was shut down on 2025-07-14, with gpt-4.1 as the recommended replacement. Source ↗

§7

Alternatives to OpenAI API

Tool From Free option Stages
Mistral API (Mistral Studio) AI model API (model maker)
Usage-based Free tier
Fireworks AI Inference provider (open-weight models)
Usage-based Trial
Together AI Inference provider (open-weight models)
Usage-based Paid
OpenRouter Model router (unified API)
— Free tier
Cerebras Inference Inference provider (open-weight models)
Usage-based Trial
Claude API AI model API (model maker)
Usage-based Trial

All AI model APIs →

§8

Questions

Does the OpenAI API have a free tier?

No free token allowance is listed on OpenAI's pricing pages; API calls are billed per token from the start. The moderation model is free.

How much does GPT-5.4 mini cost on the OpenAI API?

On 2026-09-29 it was $0.75 per 1M input tokens, $0.075 per 1M cached input tokens and $4.50 per 1M output tokens at standard rates, half that with Batch or Flex.

How do I limit how much I spend on the OpenAI API?

Each usage tier has a monthly spend ceiling, starting at $100 a month; the limit rises automatically as your cumulative payments move you up a tier.

Sources read for this page