#What Is New InAstra-6 Or Any Other Astra Version?
21 messages · Page 1 of 1 (latest)
Hi @dawn agate. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
From the knowledge I have, the only Astra model that’s documented is GPT-6 Astra (sometimes written “GPT‑6-Astra”), and there’s no information about earlier or other Astra versions. I’ll summarize what’s known about GPT‑6 Astra specifically.
What GPT‑6 Astra is
GPT‑6 Astra is described as OpenAI’s most capable model for complex work across code, apps, and research. It focuses on:
- Advanced reasoning and multi-step problem solving
- Stronger judgment and checking its own work
- Deep integration with tools and computer use [Recommended models; Introducing GPT‑6 Astra]
Key capabilities
According to the docs, GPT‑6 Astra can:
-
Coding and debugging
- Handle difficult bugs and unfamiliar problems
- Work across codebases via Codex and IDE/CLI integrations [Managing Astra usage; Recommended models]
-
Research and analysis
- Do complex research tasks, including web and file-based work (when used in ChatGPT Work / Codex with the right permissions)
- Support “Deep Research” workflows that gather information and produce cited documents, presentations, or spreadsheets [Introducing GPT‑6 Astra; Deep research]
-
Document, spreadsheet, and presentation creation
- Generate and update documents, spreadsheets, and slide decks that follow your templates and instructions
- Adapt when you add new requirements or change direction mid-task [Introducing GPT‑6 Astra; What’s new]
-
Computer use and app workflows
- Use apps and tools (via Codex / ChatGPT Work) to carry out multi-step workflows, check results, and then produce final deliverables [What’s new; Recommended models]
-
Voice and harder questions
- In ChatGPT Voice, Astra can be selected as the model used when the system needs to search or reason through harder questions, with adjustable reasoning effort [ChatGPT Voice update; Enterprise/Edu Voice]
Reasoning levels and usage
When you use GPT‑6 Astra in Work or Codex, you can choose a reasoning level (lower, medium, higher effort). Higher effort means more extensive reasoning and potentially better handling of very hard problems, but it also uses more of your allowance. Astra at low effort can still outperform GPT‑5.6 Sol at high effort in many cases [Managing Astra usage].
Availability and access
-
GPT‑6 Astra is available in:
- ChatGPT Work and Codex (desktop, web, some enterprise environments) [What’s new]
- The Codex CLI and some cloud catalogs (like Amazon Bedrock) [Other improvements]
- OpenAI’s official SDKs (Python, Node, Go, Java) as a model you can call via the API [Python changelog; Node changelog]
-
Rollout is staged: initially to a limited set of organizations, then broader availability depending on plan and workspace settings [Introducing GPT‑6 Astra; Recommended models].
About “Astra V6” or other Astra versions
The sources only mention GPT‑6 Astra and do not describe any separate “Astra V6”, “Astra V5”, or other Astra variants. I do not have information about other Astra versions beyond this single GPT‑6 Astra model.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
@dawn agate Kapa.ai is still learning and improving, please let me know how I did by reacting below.
can it do any editing? @fair badge
Let me check the knowledge base for you again. One moment! :books:
Yes. GPT‑6 Astra is explicitly designed to create and edit structured work products, not just generate them once.
From the docs, Astra can:
-
Edit documents, spreadsheets, and presentations
It can update existing files to follow your templates and new instructions, and adapt when you change direction mid‑task. [Introducing GPT‑6 Astra; What’s new] -
Iterate on complex work
In ChatGPT Work and Codex, you can have Astra revise drafts, refine analyses, and adjust outputs as you add requirements or corrections. [Introducing GPT‑6 Astra] -
Edit code and workflows
It can refactor and fix code, and modify multi-step workflows across apps and tools, not just write them once. [What’s new]
The exact editing UI depends on where you use it (ChatGPT Work, Codex desktop, IDE/CLI), but the model itself is built to support iterative editing and refinement.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
@dawn agate Kapa.ai is still learning and improving, please let me know how I did by reacting below.
i mean like Capcut or any other platform, can it edit any videos in Capcut or something? and how? @fair badge
Let me check the knowledge base for you again. One moment! :books:
There’s no information in the provided sources that GPT‑6 Astra (or any Astra model) can directly control or edit videos inside CapCut or similar third‑party editors.
What is documented about video and editing is:
-
OpenAI has a Videos API (Sora) that can:
- Edit an existing video by applying a text prompt to change things like color palette or add elements, via
POST /v1/videos/edits. [Edit existing videos; CreateVideoEdit] - Extend a completed video (continue the scene) via
POST /v1/videos/extensions. [Extend completed videos; CreateVideoExtend] - These are used through code or SDKs (Python, Node, Go), not through CapCut’s UI. [Python videos; Node videos]
- Edit an existing video by applying a text prompt to change things like color palette or add elements, via
-
ChatGPT itself can accept video uploads and analyze them (describe, summarize, answer questions), but the docs only mention analysis, not actually editing the video file like CapCut does. [ChatGPT image inputs FAQ]
There is no mention of:
- Direct integration between Astra and CapCut.
- Astra pressing buttons or operating CapCut’s timeline/effects.
- A workflow where Astra edits a CapCut project file.
So, based on the sources:
- Astra can help you plan edits (e.g., suggest cuts, effects, scripts, shot lists).
- Actual video editing like in CapCut would need to be done either:
- Programmatically via the OpenAI Videos API (Sora) in your own app or pipeline, not in CapCut. [Video generation with Sora]
- Or manually by you in CapCut, following Astra’s instructions.
If you want, I can outline how to use the Videos API to do CapCut‑style edits (color changes, adding elements, extending clips) in code, but I can’t show a way for Astra to directly operate CapCut itself, because that’s not described in the docs.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
@dawn agate Kapa.ai is still learning and improving, please let me know how I did by reacting below.
@fair badge i have some clips of videos and i added them together in a capcut project, how can Astra 6 help me with it? like it puts motions and edits?
Let me check the knowledge base for you again. One moment! :books:
From the docs we have, Astra itself is not described as being able to directly control CapCut or apply effects inside your CapCut project. There’s no mention of Astra pressing buttons, adding transitions, or placing keyframes in CapCut.
However, Astra (and OpenAI tools in general) can still help your CapCut project in indirect but useful ways:
-
Plan the edit and motion design (conceptual help)
You can describe your clips and what you want (pace, mood, style, target platform), and Astra can:- Suggest where to cut or trim each clip.
- Propose transitions (cuts, fades, zooms, speed ramps) between specific clips.
- Suggest camera motions or text animations you can then recreate in CapCut.
This is “editing guidance”, not automatic editing.
-
Script, captions, and overlays
Astra can:- Write or refine your voiceover script.
- Generate on-screen text, titles, and lower-thirds.
- Create caption text you can paste into CapCut and sync manually.
-
Color and style direction
Using the same kind of language used in the Sora video-editing docs, Astra can help you define prompts like:- “Shift the color palette to teal, sand, and rust, with a warm backlight.”
- “Same shot, but moodier lighting and more contrast.”
You then translate that into CapCut’s color filters, curves, and effects yourself. The Sora docs show how text instructions can drive color/style changes in video; in CapCut you’d just apply those ideas manually. [Edit existing videos]
- If you use OpenAI’s Videos API (outside CapCut)
Separately from CapCut, OpenAI’s Sora-based Videos API can:- Edit an existing video with a text prompt (
POST /v1/videos/edits). - Extend a clip (
POST /v1/videos/extensions). [Edit existing videos; Extend completed videos]
That happens in code (Python/Node/etc.), not inside CapCut. You could: - Export your CapCut clips,
- Run them through the Videos API for AI-style edits,
- Then re-import the results into CapCut.
- Edit an existing video with a text prompt (
Because the sources do not mention any Astra–CapCut integration, Astra cannot automatically “put motions and edits” into your CapCut project. It can:
- Help you decide what motions and edits to use.
- Write scripts, captions, and style directions.
- Optionally help you use OpenAI’s own Videos API for AI-driven edits outside CapCut.
If you tell me the style you’re going for (e.g., TikTok montage, cinematic vlog, gaming edit), I can give you a concrete, step-by-step edit plan you can follow in CapCut.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
@dawn agate Kapa.ai is still learning and improving, please let me know how I did by reacting below.
Copy this prompt into any ChatGPT desktop task
Download and read https://sf16-sg.tiktokcdn.com/obj/eden-sg/962370eh7nupenuhbo/capcut_codex_plugin/SKILL.md, then follow it exactly to install and authorize the CapCut plugin.
https://github.com/Atx-Guy/capcut-mcp-server
Depending on how you prefer to do it