DeepSearch API
Explore the API. Make a real request. Use the code in your application.
Stream response
streamReceive reasoning and answers as they are generated.
Reasoning effort
reasoning_effortBalance research depth and response time.
Maximum output tokens
max_tokensLimit the length of the answer.
Research token budget
budget_tokensMaximum reasoning budget for this request.
Always search
no_direct_answerRequire research instead of answering from model memory.
curl 'https://chatsax.com/v1/deepsearch' \
-X 'POST' \
-H 'Content-Type: application/json' \
-H 'Accept: text/event-stream' \
-H "Authorization: Bearer $CHATSAX_API_KEY" \
--data-binary '{
"model": "jina-deepsearch-v1",
"messages": [
{
"role": "user",
"content": "What are the practical differences between dense retrieval and reranking? Cite your sources."
}
],
"stream": true,
"reasoning_effort": "low",
"max_tokens": 1000,
"no_direct_answer": false
}'Not connectedResearch until the evidence connects.
Combine searching, reading, and reasoning into a question-driven workflow with cited output.
Iterative research
Let the upstream research service revisit queries and inspect sources.
Visible sources
Keep citations and visited URLs alongside the answer.
Explicit effort
Control reasoning effort, budgets, and source-domain preferences.
From playground to production.
Use the same request contract in the browser, an HTTP client, or your agent tools. A Chatsax key keeps access and usage separate from the upstream provider account.
Usage & access
Know what your requests use.
Chatsax currently uses invitation-based access. Request limits are assigned to each key; provider usage is recorded when available. No public checkout or standard price plan is offered here.
Your limits
View the current key’s scopes, minute limit, and daily request limit in the dashboard.
Recorded usage
Inspect requests, response status, latency, and available token counts.
Provider costs
Different services count usage differently. A cancelled or long-running request may still consume upstream resources.
Frequently asked questions
Should I enable streaming?
Yes for long HTTP requests. Your client should process events as they arrive and support cancellation.
Are grounded answers always correct?
No. Inspect the original sources, publication dates, and uncertainty before relying on an answer.
Where is Grounding?
Jina moved the older Grounding service into DeepSearch. Structured output can express true, false, or unknown with explanations.
Is Chatsax operated by Jina AI?
Chatsax is independently operated. Jina supplies the current underlying APIs; model names and source links identify that provider.