OpenAI

Summary

OpenAI provides APIs and tools for accessing advanced language models like GPT for building AI-powered applications.

Release history

1. snapshot-20260831 seen Aug 31, 2026 · issue 012
   • ›Adds hard monthly spend limits at the organization and project level that cause API requests to return a 429 error once tracked spend reaches the configured cap.
2. snapshot-20260829 seen Aug 29, 2026 · issue 010

   OpenAI API platform adds hard spend limits for organizations and projects with 429 enforcement and spend alerts.

   • ›Adds hard monthly spend limits for organizations and projects on the OpenAI API platform; once tracked spend reaches the cap, affected API calls return a 429 error.
   • ›Adds spend alerts to notify teams before a hard limit is hit and traffic is interrupted.
3. snapshot-20260828 seen Aug 28, 2026 · issue 009

   GPT-5.6 family launches with Programmatic Tool Calling, prompt caching controls, persisted reasoning, and multi-agent orchestration in beta.

   └──▷ TRY IT
   Send a full-resolution screenshot to GPT-5.6 for analysis without downscaling, using the original image detail setting.
   copy

   $ curl https://api.openai.com/v1/chat/completions \
     -H 'Authorization: Bearer $OPENAI_API_KEY' \
     -H 'Content-Type: application/json' \
     -d '{"model": "gpt-5.6", "messages": [{"role": "user", "content": [{"type": "image_url", "image_url": {"url": "https://example.com/screenshot.png", "detail": "original"}}]}]}'

   • ›New gpt-5.6-sol (routed via gpt-5.6 alias), gpt-5.6-terra, and gpt-5.6-luna models cover high-intelligence, balanced, and high-volume workload tiers respectively.
   • ›Adds Programmatic Tool Calling to GPT-5.6, enabling structured, API-driven tool invocation.
   • ›Adds explicit prompt caching controls to GPT-5.6.
   • ›Multi-agent orchestration support in beta for GPT-5.6.
   • ›GPT-5.6 now accepts images at their original dimensions with original or auto image detail settings.
4. snapshot-20260826 seen Aug 26, 2026 · issue 008
   └──▷ TRY IT
   Pass a full-resolution image with original detail to preserve fidelity when analysing diagrams or screenshots.
   copy

   $ curl https://api.openai.com/v1/responses \
     -H 'Authorization: Bearer $OPENAI_API_KEY' \
     -H 'Content-Type: application/json' \
     -d '{"model": "gpt-5.6", "input": [{"type": "image_url", "image_url": {"url": "https://example.com/diagram.png", "detail": "original"}}]}'

   • ›Adds the gpt-5.6 model alias, which routes requests to gpt-5.6-sol by default.
   • ›Adds three new models: gpt-5.6-sol (frontier capability), gpt-5.6-terra (intelligence/cost balance), and gpt-5.6-luna (efficient high-volume workloads).
   • ›Adds persisted reasoning with max effort support in GPT-5.6.
   • ›Adds Pro mode to GPT-5.6.
   • ›Adds Multi-agent orchestration in beta via the Responses API for GPT-5.6.
5. snapshot-20260825 seen Aug 25, 2026 · issue 007

   OpenAI API adds hard monthly spend limits for organizations and projects, returning 429 errors when caps are reached.

   • ›Adds hard spend limits at the organization and project level that trigger HTTP 429 errors on API calls once tracked monthly spend reaches the configured cap.
   • ›Adds spend alerts to notify teams before a spend limit is hit and traffic is interrupted.
6. snapshot-20260824 seen Aug 24, 2026 · issue 006

   GPT-5.6 family launches with three tiers, Programmatic Tool Calling, prompt caching controls, persisted reasoning, and multi-agent orchestration beta.

   • ›Adds the gpt-5.6-sol model (frontier capability), gpt-5.6-terra (balanced intelligence/cost), and gpt-5.6-luna (efficient high-volume) — and the gpt-5.6 alias, which routes to gpt-5.6-sol.
   • ›Adds Programmatic Tool Calling to GPT-5.6, enabling structured, code-driven invocation of tools via the API.
   • ›Adds explicit prompt caching controls to GPT-5.6, giving callers direct influence over cache behavior rather than relying on automatic caching.
   • ›Adds Pro mode to GPT-5.6 for enhanced capability on demanding workloads.
   • ›Adds Multi-agent orchestration in beta to the Responses API under GPT-5.6, enabling coordinated multi-agent workflows.
   +1 moreshow less

   • ›GPT-5.6 now accepts images at their original dimensions, with original or auto image detail settings.
7. snapshot-20260823 seen Aug 23, 2026 · issue 005

   OpenAI releases GPT-Realtime-2.1 and GPT-Realtime-2.1 mini: updated realtime reasoning models with improved voice handling.

   • ›New gpt-realtime-2.1 model adds improved alphanumeric recognition, silence and noise handling, and interruption behavior for realtime reasoning workloads.
   • ›New gpt-realtime-2.1-mini model offers a faster, lower-cost distilled reasoning option for latency-sensitive or cost-constrained realtime applications.
8. snapshot-20260822 seen Aug 22, 2026 · issue 004

   OpenAI API adds per-request regional processing selection via prefixed domain for Global geography projects.

   • ›Enables per-request regional processing by using a prefixed domain with an API key from a project configured with Global geography, without changing existing eligibility, data retention, endpoint, or model support requirements.
9. snapshot-20260821 seen Aug 21, 2026 · issue 003

   OpenAI adds a Prompt Caching dashboard and transparent background support for gpt-image-2 image generation.

   └──▷ TRY IT
   Generate a product image with a transparent background for compositing — useful when you need to layer the result over a custom background without post-processing.
   copy

   $ curl https://api.openai.com/v1/images/generations \
     -H 'Authorization: Bearer $OPENAI_API_KEY' \
     -H 'Content-Type: application/json' \
     -d '{
       "model": "gpt-image-2",
       "prompt": "A sleek wireless headphone on a clean surface",
       "background": "transparent",
       "output_format": "png"
     }'

   • ›Adds background: transparent support (preview) for gpt-image-2 and gpt-image-2-2026-04-21 in the v1/images/generations, v1/images/edits, and v1/responses endpoints; output format must be png or webp (jpeg does not support transparent backgrounds).
   • ›Releases a Prompt Caching dashboard on the OpenAI API platform showing cache hit rate over time, cache reads per write, and a breakdown of cache-read, cache-write, and uncached tokens; filterable by model and service tier.
10. snapshot-20260821 seen Aug 21, 2026 · issue 001

    OpenAI adds a Prompt Caching dashboard and transparent background support for gpt-image-2 image generation.

    └──▷ TRY IT
    Generate a product image with a transparent background for compositing — useful when you need to layer the result over a custom background without post-processing.
    copy

    $ curl https://api.openai.com/v1/images/generations \
      -H 'Authorization: Bearer $OPENAI_API_KEY' \
      -H 'Content-Type: application/json' \
      -d '{
        "model": "gpt-image-2",
        "prompt": "A sleek wireless headphone on a clean surface",
        "background": "transparent",
        "output_format": "png"
      }'

    • ›Adds background: transparent support (preview) for gpt-image-2 and gpt-image-2-2026-04-21 in the v1/images/generations, v1/images/edits, and v1/responses endpoints; output format must be png or webp (jpeg does not support transparent backgrounds).
    • ›Releases a Prompt Caching dashboard on the OpenAI API platform showing cache hit rate over time, cache reads per write, and a breakdown of cache-read, cache-write, and uncached tokens; filterable by model and service tier.
11. snapshot-20260820 seen Aug 20, 2026 · issue 002

    OpenAI adds Daybreak Blue/Red security tiers with GPT-5.6 Cyber access and Ultrafast mode for GPT-5.6 Sol.

    └──▷ TRY IT
    Run a secure code review against a file using the Daybreak Blue tier for defensive security work.
    copy

    $ curl https://api.openai.com/v1/responses \
      -H 'Authorization: Bearer $OPENAI_API_KEY' \
      -H 'Content-Type: application/json' \
      -d '{
        "model": "daybreak-blue-latest",
        "input": "Review the following code for vulnerabilities and suggest patches: <paste code here>"
      }'

    Validate an exploit or run authorized penetration test analysis using the separately approved Daybreak Red tier with GPT-5.6 Cyber.
    copy

    $ curl https://api.openai.com/v1/responses \
      -H 'Authorization: Bearer $OPENAI_API_KEY' \
      -H 'Content-Type: application/json' \
      -d '{
        "model": "daybreak-red-latest",
        "input": "Analyze this proof-of-concept for exploit validity in the context of our authorized engagement: <paste PoC here>"
      }'

    • ›Adds daybreak-blue-latest model identifier via v1/responses, providing approved defenders access to general-purpose models (including gpt-5.6-sol) for vulnerability discovery, secure code review, detection engineering, incident response, malware analysis, and patch validation.
    • ›Adds daybreak-red-latest model identifier via v1/responses, providing separately approved access to gpt-5.6-cyber — a purpose-trained model for authorized vulnerability reproduction, exploit validation, penetration testing, red teaming, and complex system analysis.
    • ›Announces Ultrafast mode, a new API service tier for gpt-5.6-sol running up to 14x faster than Standard processing (limited preview).
12. snapshot-20260820 seen Aug 20, 2026 · issue 001

    OpenAI adds Daybreak Blue/Red security tiers with GPT-5.6 Cyber access and Ultrafast mode for GPT-5.6 Sol.

    └──▷ TRY IT
    Run a secure code review against a file using the Daybreak Blue tier for defensive security work.
    copy

    $ curl https://api.openai.com/v1/responses \
      -H 'Authorization: Bearer $OPENAI_API_KEY' \
      -H 'Content-Type: application/json' \
      -d '{
        "model": "daybreak-blue-latest",
        "input": "Review the following code for vulnerabilities and suggest patches: <paste code here>"
      }'

    Validate an exploit or run authorized penetration test analysis using the separately approved Daybreak Red tier with GPT-5.6 Cyber.
    copy

    $ curl https://api.openai.com/v1/responses \
      -H 'Authorization: Bearer $OPENAI_API_KEY' \
      -H 'Content-Type: application/json' \
      -d '{
        "model": "daybreak-red-latest",
        "input": "Analyze this proof-of-concept for exploit validity in the context of our authorized engagement: <paste PoC here>"
      }'

    • ›Adds daybreak-blue-latest model identifier via v1/responses, providing approved defenders access to general-purpose models (including gpt-5.6-sol) for vulnerability discovery, secure code review, detection engineering, incident response, malware analysis, and patch validation.
    • ›Adds daybreak-red-latest model identifier via v1/responses, providing separately approved access to gpt-5.6-cyber — a purpose-trained model for authorized vulnerability reproduction, exploit validation, penetration testing, red teaming, and complex system analysis.
    • ›Announces Ultrafast mode, a new API service tier for gpt-5.6-sol running up to 14x faster than Standard processing (limited preview).