# GPT-5.6

GPT-5.6 · Ikke lenger på aktiv radar. Lansering eller navneendring er ikke bekreftet.

- Utvikler: OpenAI
- Id: gpt-5.6
- Ifølge rapporter: 2026-06-25T22:41:38.000Z
- Observasjon: 2026-10-07T19:45:27.953Z
  - [dokumentasjonslenke](https://models.dev/#gpt-5.6)
- Status: Observasjon
- Ikke lenger på aktiv radar. Lansering eller navneendring er ikke bekreftet.
- Kanonisk: https://alexgetman.com/nb/models/openai/gpt-5-6/

## Spørsmål

### Når ble GPT-5.6 lansert?

Lanseringsdatoen er ikke fastslått. Signal Forge registrerte modellen første gang 2026-10-07.

## Tidslinje

- 2026-06-25T22:41:38.000Z · [Alex Getman](https://alexgetman.com/5/gpt-5-6-delayed-at-trump-s-request/)
> GPT-5.6 delayed at Trump’s request?
> 
> The Trump administration asked OpenAI to release its new model gradually over security concerns, The Information reports.
- 2026-06-26T19:21:53.000Z · [Alex Getman](https://alexgetman.com/9/feedback-on-gpt-5-6-from-the-founder-of-opencode/)
> Feedback on GPT-5.6 from the founder of OpenCode
> 
> • Dax says he usually doesn’t hype new models. But GPT-5.6 has noticeably changed how his team works.
> • The key point: he doesn’t think the model writes good code on its own. What he likes is not the out-of-the-box result, but the process of getting to good code with it.
> • According to him, GPT-5.6 understands the task better and loses fewer details along the way. It also catches things the developer may have forgotten.
> • The problems are still there: the model can still be overly defensive and write unnecessary code. Sometimes they do another pass afterward and delete half of what it wrote.
> • The most telling fact: his team’s token usage is 
- 2026-06-26T19:45:06.000Z · [Alex Getman](https://alexgetman.com/10/the-most-powerful-ai-models-are-starting-to-look-less-like-products-and-more-like-passes-into-a/)
> The most powerful AI models are starting to look less like products and more like passes into a closed club.
> 
> Publicly, we hear: “safety,” “staged rollout,” “available soon.”
> 
> In practice: some people get GPT-5.6 now, while everyone else waits for permission.
> 
> That is a dangerous precedent.
> 
> Because the real divide won’t be between people who use AI and people who don’t.
> 
> It will be between those who got the best AI in time — and those who got it after the game had already started.
- 2026-07-08T07:44:34.000Z · [Alex Getman](https://alexgetman.com/43/gpt-5-6-drops-thursday/)
> ⚡ GPT-5.6 DROPS THURSDAY
- 2026-07-30T17:51:58.749Z · [Alex Getman](https://alexgetman.com/132/openai-just-slashed-gpt-5-6-prices/)
> ⚡️ OpenAI just slashed GPT 5.6 prices
> 
> Luna is now 80% cheaper, while Terra dropped 20%. The cuts also apply to subscriptions, with both models now using less of your quota.
> 
> Luna is now cheaper than DeepSeek V4 Pro on input. One million tokens cost $0.20 vs $0.435.
- 2026-10-07T19:47:46.396Z · [OpenAI · API changelog](https://developers.openai.com/api/docs/changelog)
> Prompt Cache Diagnostics is now generally available in the Responses API for GPT-5.6 and later supported models
- 2026-11-10T17:58:10.980Z · [GitHub · openai/openai-agents-js · code](https://github.com/openai/openai-agents-js/blob/7589b9c40120b89a7d2d28a5a60b4defe1ed7bf5/packages/agents-core/src/sandbox/capabilities/compaction.ts)
> withoutPinnedDate = normalized.replace(/-\d{4}-\d{2}-\d{2}$/u, '');
> 
>   return [...new Set([normalized, withoutPinnedDate])];
> }
> 
> const SANDBOX_CONTEXT_WINDOWS_1M = new Set([
>   'gpt-6-astra',
>   'gpt-6-sol',
>   'gpt-6-luna',
>   'gpt-6.1-sol',
>   'gpt-5.6',
>   'gpt-5.6-sol',
>   'gpt-5.6-terra',
>   'gpt-5.6-luna',
>   'gpt-5.4',
>   'gpt-5.4-2026-03-05',
>   'gpt-5.4-pro',
>   'gpt-5.4-pro-2026-03-05',
>   'gpt-4.1',
>   'gpt-4.1-2025-04-14',
>   'gpt-4.1-mini',
>   'gpt-4.1-mini-2025-04-14',
>   'gpt-4.1-nano',
>   'gpt-4.1-nano-2025-04-14',
> ]);
> 
> const SANDBOX_CONTEXT_WINDOWS_400K = new Set([
>   'gpt-5',
>   'gpt-5-codex',
>   'gpt-5-mini',
>   'gpt-5-nano',
>   'gpt-5-pro',
>   'gpt-5.1',
>   'gpt-5.1-codex',
>   'gpt-5.1-codex-max',
>   'gpt-5.1-codex-mini',
>   'gpt-5.2',
>   'gpt-5.2-codex',
>   'gpt-5.2-codex-cyber',
>   'gpt-5.2-pro',
>   'gpt-5.3-codex',
>   'gpt-5.3-codex-spark',
>   'gpt-5.4-mini',
>   'gpt-5.4-nano',
> ]);
> 
> const
- 2026-11-10T17:58:16.002Z · [GitHub · openai/openai-python · code](https://github.com/openai/openai-python/blob/0ebd00647f6088fb100b03440c6c4554bbab549d/examples/module_client.py)
> nfigured just like the `OpenAI` instantiation counterpart
> openai.base_url = "https://..."
> openai.default_headers = {"x-foo": "true"}
> 
> # all API calls work in the exact same fashion as well
> stream = openai.chat.completions.create(
>     model="gpt-5.6",
>     messages=[
>         {
>             "role": "user",
>             "content": "How do I output all files in a directory using Python?",
>         },
>     ],
>     stream=True,
> )
> 
> for chunk in stream:
>     print(chunk.choices[0].delta.content or "", end="", flush=True)
> 
> print()
> 
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
- 2026-11-10T17:59:50.472Z · [GitHub · openai/openai-agents-python · code](https://github.com/openai/openai-agents-python/blob/d68961657843c7d81dfc25834484f8aac2957eb3/docs/ja/models/index.md)
> # It's also fine to pass a GPT-5 model name explicitly:
>     model="gpt-5.6-sol",
>     model_settings=ModelSettings(reasoning=Reasoning(effort="high"), verbosity="low"),
> )
> ```
> 
> レイテンシーを抑えるには、GPT-5 モデルで `reasoning.effort="none"` を使用することを推奨します。
> 
> GPT-5.6 は、既存の `reasoning` 設定を通じて、推論モード、会話ターン間で引き継がれる推論コンテキスト、`"max"` 労力レベルもサポートします。これらの制御は Responses API 経路で利用できます。
> 
> ```python
> from openai.types.shared import Reasoning
> from agents import Agent, ModelSettings
> 
> agent = Agent(
>     name="Deep research agent",
>     model="gpt-5.6-sol",
>     model_settings=ModelSettings(
>         reasoning=Reasoning(
>             mode="pro",
>             effort="max",
>             context="all_turns",
>         ),
>     ),
> )
> ```
> 
> `reasoning.mode` と `reasoning.context` は Responses 専用の設定です。Chat Completions では `reasoning.effort` のみが使用され、サポートされる労力レベルはモデルと API サーフェスによって異なります。GPT-5.6 の `"max"` 労力には Responses API を使用してください。Chat
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
- 2026-11-10T18:00:44.369Z · [GitHub · openai/codex · code](https://github.com/openai/codex/blob/82e6bacb9c6950d3b56cb153847fa80bdd698053/codex-rs/model-provider/src/amazon_bedrock/catalog.rs)
>                 "GPT-6 Luna",
>                 /*priority*/ 4,
>             ),
>             bedrock_model(
>                 bundled_openai_model(GPT_5_6_SOL_OPENAI_MODEL_ID),
>                 AMAZON_BEDROCK_GPT_5_6_SOL_MODEL_ID,
>                 "GPT-5.6 Sol",
>                 /*priority*/ 5,
>             ),
>             bedrock_model(
>                 bundled_openai_model(GPT_5_6_TERRA_OPENAI_MODEL_ID),
>                 AMAZON_BEDROCK_GPT_5_6_TERRA_MODEL_ID,
>                 "GPT-5.6 Terra",
>                 /*priority*/ 6,
>             ),
>             bedrock_model(
>                 bundled_openai_model(GPT_5_6_LUNA_OPENAI_MODEL_ID),
>                 AMAZON_BEDROCK_GPT_5_6_LUNA_MODEL_ID,
>                 "GPT-5.6 Luna",
>                 /*priority*/ 7,
>             ),
>             gpt_5_bedrock_model(
>                 GPT_5_5_OPENAI_MODEL_ID,
>                 AMAZON_BEDROCK_GPT_5_5_MODEL_ID,
>     
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
- 2026-11-10T18:00:44.369Z · [GitHub · openai/codex · code](https://github.com/openai/codex/blob/627b579219d492b5445d6d433d3110eb53e507dc/codex-rs/backend-client/src/client/chatgpt_turn_cost_tests.rs)
> _body_json(serde_json::json!({
>             "threads": [
>                 {"thread_id": "thread-a", "turns": [{
>                     "turn_id": "turn-a", "model": "gpt-5.6",
>                     "estimated_usage_usd_micros": 1250001,
>                     "settled_response_ids": ["resp-a", "resp-b"]
>                 }]},
>                 {"thread_id": "thread-b", "turns": [
>                     {"turn_id": "turn-b", "estimated_usage_usd_micros": null,
>                      "settled_response_id
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
- 2026-11-10T20:40:17.231Z · [GitHub · openai/openai-agents-python · code](https://github.com/openai/openai-agents-python/blob/d68961657843c7d81dfc25834484f8aac2957eb3/src/agents/sandbox/capabilities/compaction.py)
> -5.4",
>             "gpt-5.4-2026-03-05",
>             "gpt-5.4-pro",
>             "gpt-5.4-pro-2026-03-05",
>             "gpt-5.5",
>             "gpt-5.5-2026-04-23",
>             "gpt-5.5-pro",
>             "gpt-5.5-pro-2026-04-23",
>             "gpt-5.6",
>             "gpt-5.6-sol",
>             "gpt-5.6-terra",
>             "gpt-5.6-luna",
>             "gpt-4.1",
>             "gpt-4.1-2025-04-14",
>             "gpt-4.1-mini",
>             "gpt-4.1-mini-2025-04-14",
>             "gpt-4.1-nano",
>             "gpt-4.1-nano-2025-04-14",
>         ),
>         1_047_576,
>     ),
>     **_model_context_windows(
>         (
>             "gpt-5",
>             "gpt-5-2025-08-07",
>             "gpt-5-codex",
>             "gpt-5-mini",
>             "gpt-5-mini-2025-08-07",
>             "gpt-5-nano",
>             "gpt-5-nano-2025-08-07",
>             "gpt-5-pro",
>             "gpt-5-pro-2025-10-06",
>             "gpt-5
- 2026-11-10T20:40:39.415Z · [GitHub · openai/openai-python · code](https://github.com/openai/openai-python/blob/0ebd00647f6088fb100b03440c6c4554bbab549d/api_reference/openapi.transformed.yml)
>       - type: string
>             description: |
>               Controls which reasoning items are rendered back to the model on later turns.
>               If omitted or set to `auto`, the model determines the context mode. The
>               `gpt-5.6` model family defaults to `all_turns`; earlier models default to
>               `current_turn`.
> 
>               When returned on a response, this is the effective reasoning context mode
>               used for the response.
>             enum:
>             - auto
>             - current_turn
>             - all_turns
>           - type: 'null'
>         generate_summary:
>           anyOf:
>           - type: string
>             deprecated: true
>             description: |
>               **Deprecated:** use `summary` instead.
> 
>               A summary of the reasoning performed by the model. This can be
>               useful for debugging and understandi
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
**Bekreftet bruk**
- 2026-11-11T09:22:02.252Z · [OpenAI · official articles · Cogent and OpenAI build cyber defense factories for enterprises](https://openai.com/index/cogent-security)
> The company’s AI agents combine GPT‑6.1 Sol and GPT‑5.6 with its enterprise context and controls to handle investigation through verified remediation
- 2026-11-12T18:45:20.401Z · [GitHub · openai/openai-node · code](https://github.com/openai/openai-node/blob/b579bcb3e586d22269ffe6fe504653a0e05acbe8/api_reference/openapi.transformed.yml)
>       - type: string
>             description: |
>               Controls which reasoning items are rendered back to the model on later turns.
>               If omitted or set to `auto`, the model determines the context mode. The
>               `gpt-5.6` model family defaults to `all_turns`; earlier models default to
>               `current_turn`.
> 
>               When returned on a response, this is the effective reasoning context mode
>               used for the response.
>             enum:
>             - auto
>             - current_turn
>             - all_turns
>           - type: 'null'
>         generate_summary:
>           anyOf:
>           - type: string
>             deprecated: true
>             description: |
>               **Deprecated:** use `summary` instead.
> 
>               A summary of the reasoning performed by the model. This can be
>               useful for debugging and understandi
Denne referansen i koden bekrefter ikke en modellidentifikator i bruk.
- 2026-11-13T18:47:23.489Z · [GitHub search · OpenAI · pulls · yc-software/qm · @ReganBell](https://github.com/yc-software/qm/pull/3579)
> sandbox and app provider, real model turns on GPT-5.6 Terra), driven through the portal in a headless Chromium at commit `31044aea1`. No apps domain is
**Bekreftet bruk**
- 2026-11-13T21:13:01.170Z · [OpenAI · official articles · Introducing the ChatGPT for small business program](https://openai.com/index/introducing-chatgpt-small-business-program)
> ChatGPT Work is also powered by GPT‑5.6, our most advanced model available to businesses of any size and all subscription plans
- 2026-11-14T06:58:53.887Z · [OpenAI · official articles](https://openai.com/index/introducing-gpt-6-cyber)
> We will progressively expand access to all Daybreak Red customers soon, and are also making GPT‑5.6 Cyber available to all Daybreak Red customers
- 2026-11-14T12:35:21.037Z · [OpenAI · official articles](https://openai.com/index/third-moment-research)
- 2026-11-14T12:55:34.251Z · [OpenAI · official articles](https://openai.com/index/pollo-ai)
**Bekreftet bruk**
- 2026-11-14T12:55:34.251Z · [OpenAI · official articles · Pollo AI turns creative ideas into campaigns with OpenAI](https://openai.com/index/pollo-ai)

## Dokumentasjonslenker

- [editorial](https://alexgetman.com/10/the-most-powerful-ai-models-are-starting-to-look-less-like-products-and-more-like-passes-into-a/)
- [editorial](https://alexgetman.com/132/openai-just-slashed-gpt-5-6-prices/)
- [editorial](https://alexgetman.com/43/gpt-5-6-drops-thursday/)
- [editorial](https://alexgetman.com/5/gpt-5-6-delayed-at-trump-s-request/)
- [editorial](https://alexgetman.com/9/feedback-on-gpt-5-6-from-the-founder-of-opencode/)
- [OpenAI · API changelog](https://developers.openai.com/api/docs/changelog)
- [GitHub · openai/codex · code](https://github.com/openai/codex/blob/627b579219d492b5445d6d433d3110eb53e507dc/codex-rs/backend-client/src/client/chatgpt_turn_cost_tests.rs)
- [GitHub · openai/codex · code](https://github.com/openai/codex/blob/82e6bacb9c6950d3b56cb153847fa80bdd698053/codex-rs/model-provider/src/amazon_bedrock/catalog.rs)
- [GitHub · openai/openai-agents-js · code](https://github.com/openai/openai-agents-js/blob/7589b9c40120b89a7d2d28a5a60b4defe1ed7bf5/packages/agents-core/src/sandbox/capabilities/compaction.ts)
- [GitHub · openai/openai-agents-python · code](https://github.com/openai/openai-agents-python/blob/d68961657843c7d81dfc25834484f8aac2957eb3/docs/ja/models/index.md)
- [GitHub · openai/openai-agents-python · code](https://github.com/openai/openai-agents-python/blob/d68961657843c7d81dfc25834484f8aac2957eb3/src/agents/sandbox/capabilities/compaction.py)
- [GitHub · openai/openai-node · code](https://github.com/openai/openai-node/blob/b579bcb3e586d22269ffe6fe504653a0e05acbe8/api_reference/openapi.transformed.yml)
- [GitHub · openai/openai-python · code](https://github.com/openai/openai-python/blob/0ebd00647f6088fb100b03440c6c4554bbab549d/api_reference/openapi.transformed.yml)
- [GitHub · openai/openai-python · code](https://github.com/openai/openai-python/blob/0ebd00647f6088fb100b03440c6c4554bbab549d/examples/module_client.py)
- [GitHub search · OpenAI · pulls · yc-software/qm · @ReganBell](https://github.com/yc-software/qm/pull/3579)
- [models-dev](https://models.dev/#gpt-5.6)
- [OpenAI · official articles · Cogent and OpenAI build cyber defense factories for enterprises](https://openai.com/index/cogent-security)
- [openai-articles](https://openai.com/index/introducing-chatgpt-small-business-program)
- [openai-articles](https://openai.com/index/introducing-gpt-6-cyber)
- [openai-articles](https://openai.com/index/pollo-ai)
- [openai-articles](https://openai.com/index/third-moment-research)

## Signaler