OpenAI

In March 2026 it closed a $122B round at an $852B post-money valuation and has filed confidentially for an IPO. It positions itself as the frontier-model platform: the Responses API is now the primary surface (Chat Completions is maintained but several new models are Responses-only), with GPT-6 Astra as the flagship and the GPT-5.6 Sol/Terra/Luna trio as the price-performance ladder. Sources: https://openai.com/index/accelerating-the-next-phase-ai/ and https://developers.openai.com/api/docs/models

A Access and limits

Sign up at platform.openai.com and create a key under your organization; auth is Authorization: Bearer $OPENAI_API_KEY. Usage tiers (https://developers.openai.com/api/docs/guides/rate-limits): Free (allowed geographies only, $100/month usage cap), Tier 1 after $5 paid ($100/mo), Tier 2 after $50 ($500/mo), Tier 3 after $100 ($1,000/mo), Tier 4 after $250 ($5,000/mo), Tier 5 after $1,000 paid ($200,000/mo). Per-model RPM/TPM caps rise with tier (e.g.

GPT-6 Astra: 500 RPM / 500K TPM at Tier 1 up to 15,000 RPM / 40M TPM at Tier 5). Service tiers change the price: Batch and Flex are 50% of Standard; Fast mode (formerly "Priority processing", renamed 2026-07-30, accepts service_tier "priority" or "fast") is 2x Standard. LONG-CONTEXT TIER: all prices recorded here are the BASE/short-context (<=272K input tokens) rate.

On the 1.05M-context models (gpt-6-astra, gpt-5.6-sol/terra/luna, gpt-5.5, gpt-5.4, gpt-5.4-pro, gpt-5.5-pro) a request whose input exceeds 272K tokens is billed at 2x input/cache and 1.5x output for the entire request. Cache writes on the newest models cost 1.25x the uncached input rate. Regional processing (data residency) endpoints add a 10% uplift for models released on or after 2026-03-05; OpenAI models served via Amazon Bedrock are billed by AWS at possibly different rates.

GPT-5.6 Sol is on promotional pricing guaranteed at least through 2026-11-21. Cybersecurity models (gpt-5.6-cyber, gpt-5.5-cyber) and the gpt-daybreak-* aliases require separate approval via the Daybreak program (https://openai.com/daybreak/). Self-serve fine-tuning is being wound down: closed to new organizations since 2026-05-07, and no new jobs for anyone after 2027-01-06.

Tools are billed on top of tokens: web search $10/1k calls (preview tool on non-reasoning models $25/1k), file search $2.50/1k calls plus $0.10/GB/day storage, hosted shell / code interpreter $0.03-$1.92 per 20-minute container session by RAM.

C Calling the API

OpenAI accepts requests in the OpenAI Chat Completions format, so any OpenAI-compatible client works by changing the base URL. Example uses gpt-4o-mini.

curl https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://api.openai.com/v1",
    api_key=os.environ["API_KEY"],
)

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

A compatibility layer is not always the provider's full API — caching, strict tool schemas and structured output are commonly unsupported there. Check the official docs before relying on an advanced feature.

D Official references

• Homepage
• API documentation
• Pricing