Documentation 2026-10-25 23:00 UTC to 2026-10-26 23:58 UTC

6 6 

7## Fast mode7## Fast mode

8 8 

9Fast mode speeds up supported models, including GPT-6.1 Sol, GPT-6 Astra,9Fast mode speeds up supported models, including GPT-6.1 Sol, GPT-6.1 Terra,

10GPT-6 Sol, and GPT-6 Luna, where available. For GPT-5.6, the speed increase10GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna, where available. For GPT-5.6, the speed increase

11is 1.5x.11is 1.5x.

12 12 

13Use `/fast` in the CLI to toggle Fast mode. Use `/statusline` to show it in the13Use `/fast` in the CLI to toggle Fast mode. Use `/statusline` to show it in the

44 44 

453. Submit your prompt453. Submit your prompt

46 46 

47 Submit your prompt, and Codex creates a Git worktree based on the branch you selected. By default, Codex works in a ["detached HEAD"](https://git-scm.com/docs/git-checkout#_detached_head).47 Submit your prompt, and Codex creates a Git worktree based on the branch you selected. By default, Codex works in a [detached HEAD](https://git-scm.com/docs/git-checkout#_detached_head) state.

48 48 

494. Choose where to keep working494. Choose where to keep working

50 50 

133 133 

134## Advanced details134## Advanced details

135 135 

136### Codex-managed and permanent worktrees136<a id="codex-managed-and-permanent-worktrees"></a>

137 137 

138By default, chats use a Codex-managed worktree. These are meant to feel lightweight and disposable. A Codex-managed worktree is typically dedicated to one chat, and Codex returns that chat to the same worktree if you hand it back there later.138### Codex-managed worktrees

139 139 

140If you want a long-lived environment, create a permanent worktree from the three-dot menu on a project in the sidebar. This creates a new permanent worktree as its own project. Permanent worktrees aren't automatically deleted, and you can start multiple chats from the same worktree.140By default, chats use a Codex-managed worktree. These are meant to feel lightweight and disposable. A Codex-managed worktree is typically dedicated to one chat, and Codex returns that chat to the same worktree if you hand it back there later.

141 141 

142### How Codex manages worktrees for you142### How Codex manages worktrees for you

143 143 

175<ToggleSection title="Why this limitation exists">175<ToggleSection title="Why this limitation exists">

176Git prevents the same branch from being checked out in more than one worktree at a time because a branch represents a single mutable reference (`refs/heads/<name>`) whose meaning is “the current checked-out state” of a working tree.176Git prevents the same branch from being checked out in more than one worktree at a time because a branch represents a single mutable reference (`refs/heads/<name>`) whose meaning is “the current checked-out state” of a working tree.

177 177 

178When a branch is checked out, Git treats its HEAD as owned by that worktree and expects operations like commits, resets, rebases, and merges to advance that reference in a well-defined, serialized way. Allowing multiple worktrees to simultaneously check out the same branch would create ambiguity and race conditions around which worktree’s operations update the branch reference, potentially leading to lost commits, inconsistent indexes, or unclear conflict resolution.178When a branch is checked out, Git treats its HEAD as owned by that worktree and expects operations like `git commit`, `git reset`, `git rebase`, and `git merge` to advance that reference in a well-defined, serialized way. Allowing multiple worktrees to simultaneously check out the same branch would create ambiguity and race conditions around which worktree’s operations update the branch reference, potentially leading to lost commits, inconsistent indexes, or unclear conflict resolution.

179 179 

180By enforcing a one-branch-per-worktree rule, Git guarantees that each branch has a single authoritative working copy, while still allowing other worktrees to safely reference the same commits via detached HEADs or separate branches.180By enforcing a one-branch-per-worktree rule, Git guarantees that each branch has a single authoritative working copy, while still allowing other worktrees to safely reference the same commits in a detached HEAD state or on separate branches.

181 181 

182</ToggleSection>182</ToggleSection>

183 183 

191 191 

192- A pinned chat is tied to it192- A pinned chat is tied to it

193- The chat is still in progress193- The chat is still in progress

194- The worktree is a permanent worktree194 

195Permanent worktrees created in earlier versions remain available and protected from automatic cleanup. You can no longer create permanent worktrees from the project menu.

195 196 

196Codex-managed worktrees are deleted automatically when:197Codex-managed worktrees are deleted automatically when:

197 198 

223 Chats can remain in your history even if the underlying worktree directory is224 Chats can remain in your history even if the underlying worktree directory is

224 deleted. For Codex-managed worktrees, Codex saves a snapshot before deleting225 deleted. For Codex-managed worktrees, Codex saves a snapshot before deleting

225 the worktree and offers to restore it if you reopen the associated chat.226 the worktree and offers to restore it if you reopen the associated chat.

226 Permanent worktrees are not automatically deleted when you archive their

227 chats.

228</ToggleSection>227</ToggleSection>

44 44 

453. Submit your prompt453. Submit your prompt

46 46 

47 Submit your prompt, and Codex creates a Git worktree based on the branch you selected. By default, Codex works in a ["detached HEAD"](https://git-scm.com/docs/git-checkout#_detached_head).47 Submit your prompt, and Codex creates a Git worktree based on the branch you selected. By default, Codex works in a [detached HEAD](https://git-scm.com/docs/git-checkout#_detached_head) state.

48 48 

494. Choose where to keep working494. Choose where to keep working

50 50 

133 133 

134## Advanced details134## Advanced details

135 135 

136### Codex-managed and permanent worktrees136<a id="codex-managed-and-permanent-worktrees"></a>

137 137 

138By default, chats use a Codex-managed worktree. These are meant to feel lightweight and disposable. A Codex-managed worktree is typically dedicated to one chat, and Codex returns that chat to the same worktree if you hand it back there later.138### Codex-managed worktrees

139 139 

140If you want a long-lived environment, create a permanent worktree from the three-dot menu on a project in the sidebar. This creates a new permanent worktree as its own project. Permanent worktrees aren't automatically deleted, and you can start multiple chats from the same worktree.140By default, chats use a Codex-managed worktree. These are meant to feel lightweight and disposable. A Codex-managed worktree is typically dedicated to one chat, and Codex returns that chat to the same worktree if you hand it back there later.

141 141 

142### How Codex manages worktrees for you142### How Codex manages worktrees for you

143 143 

175<ToggleSection title="Why this limitation exists">175<ToggleSection title="Why this limitation exists">

176Git prevents the same branch from being checked out in more than one worktree at a time because a branch represents a single mutable reference (`refs/heads/<name>`) whose meaning is “the current checked-out state” of a working tree.176Git prevents the same branch from being checked out in more than one worktree at a time because a branch represents a single mutable reference (`refs/heads/<name>`) whose meaning is “the current checked-out state” of a working tree.

177 177 

178When a branch is checked out, Git treats its HEAD as owned by that worktree and expects operations like commits, resets, rebases, and merges to advance that reference in a well-defined, serialized way. Allowing multiple worktrees to simultaneously check out the same branch would create ambiguity and race conditions around which worktree’s operations update the branch reference, potentially leading to lost commits, inconsistent indexes, or unclear conflict resolution.178When a branch is checked out, Git treats its HEAD as owned by that worktree and expects operations like `git commit`, `git reset`, `git rebase`, and `git merge` to advance that reference in a well-defined, serialized way. Allowing multiple worktrees to simultaneously check out the same branch would create ambiguity and race conditions around which worktree’s operations update the branch reference, potentially leading to lost commits, inconsistent indexes, or unclear conflict resolution.

179 179 

180By enforcing a one-branch-per-worktree rule, Git guarantees that each branch has a single authoritative working copy, while still allowing other worktrees to safely reference the same commits via detached HEADs or separate branches.180By enforcing a one-branch-per-worktree rule, Git guarantees that each branch has a single authoritative working copy, while still allowing other worktrees to safely reference the same commits in a detached HEAD state or on separate branches.

181 181 

182</ToggleSection>182</ToggleSection>

183 183 

191 191 

192- A pinned chat is tied to it192- A pinned chat is tied to it

193- The chat is still in progress193- The chat is still in progress

194- The worktree is a permanent worktree194 

195Permanent worktrees created in earlier versions remain available and protected from automatic cleanup. You can no longer create permanent worktrees from the project menu.

195 196 

196Codex-managed worktrees are deleted automatically when:197Codex-managed worktrees are deleted automatically when:

197 198 

223 Chats can remain in your history even if the underlying worktree directory is224 Chats can remain in your history even if the underlying worktree directory is

224 deleted. For Codex-managed worktrees, Codex saves a snapshot before deleting225 deleted. For Codex-managed worktrees, Codex saves a snapshot before deleting

225 the worktree and offers to restore it if you reopen the associated chat.226 the worktree and offers to restore it if you reopen the associated chat.

226 Permanent worktrees are not automatically deleted when you archive their

227 chats.

228</ToggleSection>227</ToggleSection>

27 27 

28## Availability and controls28## Availability and controls

29 29 

30Composer predictions is currently in beta for users aged 18 and older on personal ChatGPT Pro plans in supported regions. It’s available in local and SSH threads in the Codex desktop app with GPT-6 Astra and GPT-6.1 Sol. Predictions are enabled by default for eligible users. You can turn them on or off using the **Show predictions** toggle in **Settings → General → Composer**.30Composer predictions is currently in beta for users aged 18 and older on personal ChatGPT Pro plans in supported regions. It’s available in local and SSH threads in the Codex desktop app with GPT-6 Astra, GPT-6.1 Sol, and GPT-6.1 Terra. Predictions are enabled by default for eligible users. You can turn them on or off using the **Show predictions** toggle in **Settings → General → Composer**.

31 31 

32Generating predictions doesn’t count toward your Codex usage limits or consume credits. When you send a message, including an accepted prediction, normal Codex usage limits and billing apply.32Generating predictions doesn’t count toward your Codex usage limits or consume credits. When you send a message, including an accepted prediction, normal Codex usage limits and billing apply.

33 33 

196 },196 },

197 {197 {

198 key: "Codex-managed worktree",198 key: "Codex-managed worktree",

199 href: "/codex/environments/git-worktrees#codex-managed-and-permanent-worktrees",199 href: "/codex/environments/git-worktrees#codex-managed-worktrees",

200 appliesTo: "Desktop app",200 appliesTo: "Desktop app",

201 description: "A temporary worktree Codex creates and manages for a chat.",201 description: "A temporary worktree Codex creates and manages for a chat.",

202 },202 },

480 description:480 description:

481 "JSON Schema passed to `codex exec` to constrain the final response.",481 "JSON Schema passed to `codex exec` to constrain the final response.",

482 },482 },

483 {

484 key: "Permanent worktree",

485 href: "/codex/environments/git-worktrees#codex-managed-and-permanent-worktrees",

486 appliesTo: "Desktop app",

487 description: "A long-lived worktree kept as its own project.",

488 },

489 {483 {

490 key: "Permission profile",484 key: "Permission profile",

491 href: "/codex/permissions#define-and-select-a-profile",485 href: "/codex/permissions#define-and-select-a-profile",

16 16 

17<ModelRecommender client:load />17<ModelRecommender client:load />

18 18 

19### When to consider GPT-6.1 Terra

20 

21Try [GPT-6.1 Terra](https://developers.openai.com/api/docs/models/gpt-6.1-terra) for concise writing based on

22supplied material, computer-use tasks with a defined outcome, and scoped code

23fixes. Provide clear inputs, requirements, and a way to verify the result, such as

24checking saved records or running regression tests.

25 

26Compare it with GPT-6.1 Sol on the same inputs and reasoning effort. For writing,

27judge audience fit, voice, and useful detail as well as accuracy and length. For

28computer use, check the saved result and the accuracy of the completion report.

29 

19### When to consider GPT-6.1 Sol30### When to consider GPT-6.1 Sol

20 31 

21Consider GPT-6.1 Sol for complex projects where cost matters, such as32Consider GPT-6.1 Sol for complex projects where cost matters, such as

87 98 

88If you can, experiment using the same inputs to compare results and keep the99If you can, experiment using the same inputs to compare results and keep the

89lightest setting that meets your quality bar.100lightest setting that meets your quality bar.

101 

102Choose acceptance criteria before comparing models:

103 

104- **Writing:** check required facts, qualifications, audience, and voice. Review

105 both outputs before deciding whether the shorter one is more useful.

106- **Computer use:** reopen the affected record or page to verify saved changes,

107 confirm that the workflow preserved unrelated data, and compare the completion

108 report with the actual result.

109- **Code and interfaces:** run regression tests. For an interactive application,

110 also inspect the rendered result on desktop and mobile.

111 

112Repeat the comparison across representative tasks. Measure completion time

113alongside quality, including time spent reviewing or repairing the result. Then

114vary reasoning effort while keeping the model and inputs fixed.

2 2 

3> For the complete documentation index, see [llms.txt](https://learn.chatgpt.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to the page URL.3> For the complete documentation index, see [llms.txt](https://learn.chatgpt.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to the page URL.

4 4 

5## GPT-5.5 retirement5Choosing a model is about finding the right balance of speed, cost, and capability

6 6for your task. Sometimes you need a quick first draft. Other times, a complex

7Effective October 14, 2026, at noon PDT (19:00 UTC), GPT-5.5 retires from7problem calls for more careful reasoning. This guide introduces our latest

8Codex for users signed in with ChatGPT. The OpenAI API and Codex sessions8models, shows where each one shines, and helps you find a starting point for

9authenticated with an API key aren't affected.9your work.

10 

11For Codex with ChatGPT sign-in, choose an available replacement:

12 

13- On Plus, Pro, Business, Enterprise, and Edu plans, choose **GPT-6.1 Sol**

14 (`gpt-6.1-sol`) when available.

15- On Free and Go plans, choose **GPT-6 Luna** (`gpt-6-luna`) in the desktop app

16 when available.

17 

18Replace `gpt-5.5` in workspace defaults, saved model settings, managed

19configurations, custom agents, scheduled tasks, and scripts that select a model.

20 10 

21See [workspace model availability](https://learn.chatgpt.com/docs/enterprise/workspace-model-availability#prepare-for-the-gpt-55-retirement)11GPT-5.5 retirement: [See replacement models and migration

22for administrator guidance.12 steps](#gpt-55-retirement).

23 13 

24<ContentModeSwitch group="codex-surface" id="app">14<ContentModeSwitch group="codex-surface" id="app">

25 15 

33In Work or Codex in the ChatGPT desktop app, use the model and reasoning control23In Work or Codex in the ChatGPT desktop app, use the model and reasoning control

34beneath the composer to choose an available model and adjust its reasoning effort.24beneath the composer to choose an available model and adjust its reasoning effort.

35 25 

36Higher reasoning effort can improve results for complex tasks, but it takes26**Reasoning effort** guides how much work the model puts into thinking through

27your request. Higher effort can improve results for complex tasks, but it takes

37longer and uses more tokens. Start with the default effort and increase it when28longer and uses more tokens. Start with the default effort and increase it when

38the task needs deeper planning or analysis.29the task needs deeper planning or analysis.

39 30 

63model and reasoning control beneath the composer to choose an available model54model and reasoning control beneath the composer to choose an available model

64and adjust its reasoning effort.55and adjust its reasoning effort.

65 56 

66Higher reasoning effort can improve results for complex tasks, but it takes57**Reasoning effort** guides how much work the model puts into thinking through

58your request. Higher effort can improve results for complex tasks, but it takes

67longer and uses more tokens. Start with the default effort and increase it when59longer and uses more tokens. Start with the default effort and increase it when

68the task needs deeper planning or analysis.60the task needs deeper planning or analysis.

69 61 

84 76 

85## Choose a model77## Choose a model

86 78 

79GPT-6.1 Terra requires Codex CLI 0.159.0 or later. To use it, run

80`codex -m gpt-6.1-terra`.

81 

87In an interactive CLI session, use `/model` to switch models or adjust82In an interactive CLI session, use `/model` to switch models or adjust

88reasoning effort. You can also choose a model when you launch Codex with83reasoning effort. You can also choose a model when you launch Codex with

89`--model` or its `-m` alias:84`--model` or its `-m` alias:

100```95```

101 96 

102 97 

103Higher reasoning effort can improve results for complex tasks, but it takes98**Reasoning effort** guides how much work the model puts into thinking through

99your request. Higher effort can improve results for complex tasks, but it takes

104longer and uses more tokens. Start with the default effort and increase it when100longer and uses more tokens. Start with the default effort and increase it when

105the task needs deeper planning or analysis.101the task needs deeper planning or analysis.

106 102 

123Use the model switcher below the composer to choose an available model and119Use the model switcher below the composer to choose an available model and

124reasoning effort.120reasoning effort.

125 121 

126Higher reasoning effort can improve results for complex tasks, but it takes122**Reasoning effort** guides how much work the model puts into thinking through

123your request. Higher effort can improve results for complex tasks, but it takes

127longer and uses more tokens. Start with the default effort and increase it when124longer and uses more tokens. Start with the default effort and increase it when

128the task needs deeper planning or analysis.125the task needs deeper planning or analysis.

129 126 

138 135 

139 136 

140 137 

138</ContentModeSwitch>

139 

140<ContentModeSwitch group="codex-surface" ids="app,web">

141 

142In the ChatGPT desktop app and ChatGPT Work on the web, start with the default

143 Power setting available to your account. Move toward **Smarter** for deeper

144 reasoning or **Faster** for faster, lower-cost work. Open **Advanced** to

145 choose a specific model, reasoning effort, or speed. To try GPT-6.1 Terra in

146 this example, open **Advanced** and choose it from the model list.

147 

148The six Power presets shown are Luna High, GPT-6 Sol Light (selected

149in this example), GPT-6 Sol Medium, Astra Light, Astra Medium, and Astra Extra

150High. Some paid plans omit Astra Extra High. Advanced controls are illustrative;

151available options and defaults vary by plan, client, workspace settings, and rollout.

152 

141</ContentModeSwitch>153</ContentModeSwitch>

142 154 

143<a id="recommended-models"></a>155<a id="recommended-models"></a>

151 163 

152## Recommended models164## Recommended models

153 165 

154For complex coding and agentic workflows, use GPT-6.1 Sol when available to your166Start with the kind of work you want to do. Astra suits the most demanding

155account and client. Use Luna for focused, repeatable tasks. Select `gpt-6.1-sol`167workflows. GPT-6.1 Sol offers near-Astra performance at a lower cost than Astra.

156or `gpt-6-luna` in your model picker or saved configuration when available.168GPT-6.1 Terra is a fast, efficient choice for writing, visual understanding, and

157 169computer use. Luna suits focused, repeatable tasks.

158GPT-6.1 Sol offers near-Astra performance for complex work at a lower cost

159than Astra. See [GPT-6.1 Sol](#gpt-61-sol) for its separate rollout.

160 170 

161In ChatGPT, GPT-6.1 Sol, GPT-6 Sol, and GPT-6 Luna are available in Work and171These recommendations apply to ChatGPT Work and Codex. For Chat, see [models in ChatGPT](https://help.openai.com/en/articles/20001354-gpt-6-and-other-models-in-chatgpt) in the Help Center.

162Codex. They aren't available in Chat.

163 

164GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna remain available during the rollout. Selecting a

165new model doesn't change workspace permissions or grant access to it.

166 

167For Ultrafast mode availability and credit usage on Pro $500 and eligible

168Enterprise and Edu plans, see [Speed](https://learn.chatgpt.com/docs/agent-configuration/speed#ultrafast-mode).

169 172 

170<a id="app-compare-models"></a>173<a id="app-compare-models"></a>

171 174 

241 }}244 }}

242/>245/>

243 246 

247 

248 

249 

250<a

251 id="gpt-61-terra"

252 data-codex-legacy-anchor

253 className="block scroll-mt-24"

254></a>

255 

256<ModelDetails

257 client:load

258 name="gpt-6.1-terra"

259 slug="gpt-6.1-terra"

260 imageLabel="6.1 Terra"

261 wallpaperUrl="/images/api/models/gpt-6.1-terra-block.webp"

262 description="A fast, efficient model for clear writing, visual understanding, and computer use. Try it for drafting proposals, interpreting charts, or working with information in an app."

263 data={{

264 features: [

265 {

266 title: "Capability",

267 value: "",

268 icons: [

269 "openai.SparklesFilled",

270 "openai.SparklesFilled",

271 "openai.SparklesFilled",

272 "openai.SparklesFilled",

273 "openai.SparklesFilled",

274 ],

275 },

276 {

277 title: "Speed",

278 value: "",

279 icons: ["openai.Flash", "openai.Flash", "openai.Flash", "openai.Flash"],

280 },

281 { title: "Work and Codex in the desktop app", value: true },

282 { title: "ChatGPT Work on the web", value: true },

283 { title: "Codex CLI", value: true },

284 { title: "ChatGPT Credits", value: true },

285 { title: "API Access", value: true },

286 ],

287 }}

288/>

289 

290 

291 

292 

244<ModelDetails293<ModelDetails

245 client:load294 client:load

246 name="gpt-6-luna"295 name="gpt-6-luna"

277 326 

278 327 

279 328 

280Availability depends on the rollout, your sign-in method, and your client.

281See [pricing](https://learn.chatgpt.com/docs/pricing) for plan access and usage, and

282[workspace model availability](https://learn.chatgpt.com/docs/enterprise/workspace-model-availability)

283for Enterprise access.

284 

285</ContentModeSwitch>329</ContentModeSwitch>

286 330 

331<a id="gpt-61-sol"></a>

287<a id="gpt-6.1-sol"></a>332<a id="gpt-6.1-sol"></a>

288 333 

289### GPT-6.1 Sol334For model availability, pricing, and usage limits, see [ChatGPT Work and Codex](https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex) in the Help Center.

290 

291The GPT-6.1 Sol launch rollout includes Plus, Pro, Business, Enterprise, and Edu

292in Codex in the desktop app and CLI, and ChatGPT Work on the web and mobile.

293For Enterprise and Edu, the plan keeps GPT-6.1 Sol off by default until an

294administrator enables it.

295Free and Go are not included at launch.

296 

297Standard and Fast modes are available where supported. Ultrafast is available

298on Pro $500 and eligible Enterprise and Edu plans. See

299[Ultrafast mode](https://learn.chatgpt.com/docs/agent-configuration/speed#ultrafast-mode) for plan and

300workspace requirements.

301 

302Use `gpt-6.1-sol`. Available controls depend on your plan, client, and workspace

303settings.

304Reasoning effort ranges from Light to Ultra. Max and Ultra depend on your settings.

305 

306See [pricing](https://learn.chatgpt.com/docs/pricing) for credit usage and

307[model selection](https://developers.openai.com/api/docs/guides/model-selection#when-to-consider-gpt-61-sol)

308for guidance on comparing it with Astra and Luna. To see how its API

309specifications differ from the previous Sol model, [compare GPT-6.1 Sol and

310GPT-6 Sol](https://developers.openai.com/api/docs/models/compare?model=gpt-6.1-sol&model2=gpt-6-sol).

311 

312In the ChatGPT desktop app and ChatGPT Work on the web, start with the default

313 Power setting available to your account. Move toward **Smarter** for deeper

314 reasoning or **Faster** for faster, lower-cost work. Open **Advanced** to

315 choose a specific model, reasoning effort, or speed.

316 335 

317<ContentModeSwitch group="codex-surface" ids="app,web,cli,ide">336<ContentModeSwitch group="codex-surface" ids="app,web,cli,ide">

318 337 

319<ContentModeSwitch group="codex-surface" ids="app,web">

320 

321The six Power presets shown are Luna High, GPT-6 Sol Light (selected

322in this example), GPT-6 Sol Medium, Astra Light, Astra Medium, and Astra Extra

323High. Some paid plans omit Astra Extra High. Advanced controls are illustrative;

324available options and defaults vary by plan, client, workspace settings, and rollout.

325 

326</ContentModeSwitch>

327 

328<a id="choosing-sol-terra-and-luna"></a>338<a id="choosing-sol-terra-and-luna"></a>

329 339 

330<a id="choosing-astra-sol-and-luna"></a>340<a id="choosing-astra-sol-and-luna"></a>

335 345 

336## Choosing a model346## Choosing a model

337 347 

338Choose **Astra** when a task needs the strongest capability across steps and348Start with a task you know well and compare the result with what you need.

339tools. **GPT-6.1 Sol** offers near-Astra performance at a lower cost349Consider how long you're willing to wait, how much usage the task requires,

340than Astra. Use **GPT-6.1 Sol** when available to your account and client, and350and the level of detail or judgment involved. You can change models as your

341**Luna** for clear, repeatable tasks.351work changes.

342 352 

343### Where each model shines353### Where each model shines

344 354 

350- **GPT-6.1 Sol, for repeated, long-running work.** Consider GPT-6.1 Sol for360- **GPT-6.1 Sol, for repeated, long-running work.** Consider GPT-6.1 Sol for

351 work across code, apps, and documents when cost matters. Keep Astra for your361 work across code, apps, and documents when cost matters. Keep Astra for your

352 most demanding work.362 most demanding work.

363- **GPT-6.1 Terra, for everyday writing and visual work.** Choose Terra for clear

364 communication, interpreting images and screens, and completing tasks across

365 supported apps. Try it for drafting proposals, reviewing visual materials,

366 or filling out forms from source documents.

353- **Luna, for clear, repeatable tasks.** Choose Luna for specific, high-volume367- **Luna, for clear, repeatable tasks.** Choose Luna for specific, high-volume

354 tasks when you know what a good result looks like, such as extraction,368 tasks when you know what a good result looks like, such as extraction,

355 classification, transformation, and structured summaries.369 classification, transformation, and structured summaries.

356 370 

371### Try an example

372 

373Use your own materials with these prompts, and tell ChatGPT what a useful

374result should include.

375 

376| Model | Task | Example prompt |

377| ------------- | ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

378| Astra | Work through a complex decision. | "Compare these three vendor proposals against our requirements. Explain the tradeoffs, recommend an option, and list the questions we should resolve before deciding." |

379| GPT-6.1 Sol | Bring several sources into a working document. | "Use this project brief, meeting notes, and timeline to draft an implementation plan. Identify dependencies, flag conflicting details, and list the decisions we still need to make." |

380| GPT-6.1 Terra | Make sense of visual information. | "Review these dashboard screenshots and summarize the main changes. Refer to the chart labels and values, and flag anything you can't read confidently." |

381| Luna | Organize information into a consistent format. | "Turn these meeting notes into a table of action items, owners, and due dates. Mark any missing owner or date as unspecified." |

382 

383Review the result against your source materials, especially names, numbers,

384and details taken from images. For tasks on websites or in desktop apps, check

385the [browser use](https://learn.chatgpt.com/docs/browser) and [computer use](https://learn.chatgpt.com/docs/computer-use) setup

386requirements first.

387 

357### Pick a reasoning effort388### Pick a reasoning effort

358 389 

390For example, you might use a lower effort to rewrite a short email and a higher

391effort to compare several proposals with competing requirements.

392 

359For GPT-6.1 Sol, start with the reasoning effort available by default in your393For GPT-6.1 Sol, start with the reasoning effort available by default in your

360client and adjust based on the task. Start with **High** for Luna or **Light**394client and adjust based on the task. Start with **High** for Luna or **Light**

361for Astra.395for Astra.

392 426 

393</ContentModeSwitch>427</ContentModeSwitch>

394 428 

429## GPT-5.5 retirement

430 

431Effective October 14, 2026, at noon PDT (19:00 UTC), GPT-5.5 retires from

432Codex for users signed in with ChatGPT. The OpenAI API and Codex sessions

433authenticated with an API key aren't affected.

434 

435For Codex with ChatGPT sign-in, choose an available replacement:

436 

437- On Plus, Pro, Business, Enterprise, and Edu plans, choose **GPT-6.1 Sol**

438 (`gpt-6.1-sol`) when available.

439- On Free and Go plans, choose **GPT-6 Luna** (`gpt-6-luna`) in the desktop app

440 when available.

441 

442Replace `gpt-5.5` in workspace defaults, saved model settings, managed

443configurations, custom agents, scheduled tasks, and scripts that select a model.

444 

445See [workspace model availability](https://learn.chatgpt.com/docs/enterprise/workspace-model-availability#prepare-for-the-gpt-55-retirement)

446for administrator guidance.

447 

395<ContentModeSwitch group="codex-surface" ids="app,cli,ide">448<ContentModeSwitch group="codex-surface" ids="app,cli,ide">

396 449 

397## Other models450## Other models

6 6 

7## Fast mode7## Fast mode

8 8 

9Fast mode speeds up supported models, including GPT-6.1 Sol, GPT-6 Astra,9Fast mode speeds up supported models, including GPT-6.1 Sol, GPT-6.1 Terra,

10GPT-6 Sol, and GPT-6 Luna, where available. For GPT-5.6, the speed increase10GPT-6 Astra, GPT-6 Sol, and GPT-6 Luna, where available. For GPT-5.6, the speed increase

11is 1.5x.11is 1.5x.

12 12 

13Use `/fast` in the CLI to toggle Fast mode. Use `/statusline` to show it in the13Use `/fast` in the CLI to toggle Fast mode. Use `/statusline` to show it in the

6 6 

7 7 

8 8 

9## [October 26–30, 2026](https://learn.chatgpt.com/docs/whats-new/october-26-30-2026)

10 

11 

12 

13 

14### Use GPT-6.1 Terra for visual reasoning and computer use

15 

16[GPT-6.1 Terra](https://learn.chatgpt.com/docs/models?surface=app#app-gpt-61-terra) is now available in ChatGPT Work

17and Codex on Plus, Pro, Business, Enterprise, and Edu plans. This fast, efficient model for professional work combines

18strong image understanding with precise computer use. Match product photos to

19catalog entries, review design assets, and act on what you find across your

20browser, desktop, and apps.

21 

22[Read the full update →](https://learn.chatgpt.com/docs/whats-new/october-26-30-2026)

23 

9## October 12–16, 202624## October 12–16, 2026

10 25 

11 26 

1# October 26–30, 2026

2 

3> For the complete documentation index, see [llms.txt](https://learn.chatgpt.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to the page URL.

4 

5[Back to What's new](https://learn.chatgpt.com/docs/whats-new#october-2630-2026)

6 

7## Use GPT-6.1 Terra for visual reasoning and computer use

8 

9[GPT-6.1 Terra](https://learn.chatgpt.com/docs/models?surface=app#app-gpt-61-terra) is now available in ChatGPT Work

10and Codex on Plus, Pro, Business, Enterprise, and Edu plans. It's a fast,

11efficient model for professional work that combines strong image understanding

12with precise computer use.

13 

14Use Terra to reason about images and screens, then take action across your

15browser, desktop, and apps. Match product photos from a supplier lookbook to

16catalog entries and update listings, or check campaign assets for cropping and text

17placement and organize the edits.

18 

19Terra also writes in a clear, direct style, putting results first and explaining

20supporting details in plain language.

21 

22Choose GPT-6.1 Terra from the model list in **Advanced** in the ChatGPT desktop

23app or ChatGPT Work on the web. Work also supports Terra on mobile. The model isn't available in Chat and does not support UAE inference residency. Enterprise access depends on workspace

24availability and administrator controls. Browser use and computer use each

25have setup and access requirements.

26 

27See [Models](https://learn.chatgpt.com/docs/models?surface=app#app-gpt-61-terra), [browser use](https://learn.chatgpt.com/docs/browser), and

28[computer use](https://learn.chatgpt.com/docs/computer-use) to get started. Read the

29[launch announcement](https://learn.chatgpt.com/docs/changelog#codex-2026-10-26-gpt-61-terra).