For a one-off research question, ask GPT-6 Astra in ChatGPT.

Why GPT-6 Astra

The reasons it wins for this job.

OpenAI names research, documents and computer use among the jobs Astra is built for. It is its most capable model, released 3 September 2026.
It shares first place on the Artificial Analysis Intelligence Index with Claude Fable 5.1, and costs noticeably less to run on that index, so it thinks as long as the question needs.
In ChatGPT it can search the web, read what it finds and work through uploaded PDFs and spreadsheets in the same conversation.
For a one-off question the interface matters: citations you can click, a thread you can push back on, and nothing to build.

How to use it

4 steps to a first result.

1Ask the real decision"Should we enter the German market this year" gets a better answer than "tell me about Germany". Say what the research is for.
2Name the sources you trustPoint it at analyst reports, filings or specific sites, and tell it which kinds of source to ignore.
3Demand citations per claimAsk for a link next to every number. Then open three at random and check them.
4Ask what would change the answerThe best follow-up: what evidence would make you conclude the opposite? It surfaces the weak points.

Prompt for ChatGPT

Start from this.

Edit the parts in capitals, then run it.

I have to decide DECISION by DATE. Research it for me.

Context: WHO WE ARE, WHAT WE SELL, WHAT WE ALREADY KNOW.

What I need:
1. The five facts that matter most for this decision, each with a source link
2. The three strongest competitors or alternatives, and what they charge or claim, with links
3. What is still uncertain, and what evidence would settle it

Use primary sources (company sites, filings, docs, reputable press) over SEO blogs. If two sources disagree, show both. Do not state a number without a link.

Nothing is sent anywhere. It copies to your clipboard.

Alternatives that also work

If GPT-6 Astra is not an option.

Claude Fable 5.1Official page →

Anthropic · closed · cost: high

Pick it when you are on Claude and the question needs long, careful reasoning. It shares the top Intelligence Index score.

Gemini 3.8 FlashOfficial page →

Google · closed · cost: low

Pick it when you want answers grounded in Google Search, or the source is a YouTube video or long recording.

Claude Opus 5Official page →

Anthropic · closed · cost: high

Pick it when the research is the first step of a document you will write in Claude afterwards.

Watch out for

Citations can point to a real page that does not say what the model claims. Opening the link is the only check that works.
Recent events lag. For anything from the last few days, ask it to search explicitly and to say the date of each source.

Doing this again and again? For research that runs every week, use Claude Opus 5 with web search. Competitor monitoring and account research run unattended, so the model has to decide what to search, read the results and stop when it has enough. Opus 5 does that well through the API's built-in web search tool, at half Fable's price.

Sources

Checked 17 September 2026. Models change monthly; we re-check this page when they do.

Picking the model is the easy part.

Wiring it into a workflow that runs every week, with evals, fallbacks and a cost you can predict, is the work. Fifteen minutes, no deck.

Book fifteen minutes →