Search built for agentsthat predict.

Start free Read the docs

Try:

2026-08-19Options imply an 8% move for Nvidia into next week's report
Options traders are pricing a swing of roughly eight percent in either direction when the chipmaker reports next week…
2026-08-18Analysts lift data-center revenue estimates ahead of Q2 print
Several brokerages raised their data-center revenue forecasts, citing hyperscaler order books that remain full through year end…
2026-08-17Hyperscaler capex guidance points to another strong quarter for AI chips
Capital-spending guidance from the largest cloud providers implies accelerator demand well above last quarter's run rate…
2026-08-14Nvidia shares slip as export-rule headlines return
Shares fell in afternoon trading after reports that export licensing for advanced accelerators could tighten again…
2026-08-12Whisper numbers drift above consensus for Nvidia Q2
Buy-side expectations have crept above the published consensus, raising the bar for a positive reaction on the print…

sample

Built for prediction agents.

LIVE

Nvidia Q2 results due after the bell2 min

Options price an 8% earnings move14 min

Analyst lifts data-center estimate1 h

Live search over news and markets

Continuously crawled news, wires and market pages, with full text and publish time.

as_of_date = 2026-08-20

2025-01today

Options imply 8% move into the print08-19

Analysts lift data-center estimates08-18

Backtest on the record as it stood

Set as_of_date and the search runs against the record as it stood that day, not a filter over today's results.

EPISODE 4,812

question: NVDA up after earnings?
search: as_of_date = 2026-08-20
prediction: p(up) = 0.64
outcome: UP +9%
reward: +0.64

Train in a world that cannot leak

Resolved questions become episodes: the agent searches as of the question date and is rewarded on the real outcome.

An RL environment for prediction agents.

The agent loops on point-in-time search, makes a prediction, and a reward model scores it against the real outcome.

Read the docs

One call. Add a date to travel.

curl https://api.ariselabs-search-api.com/v1/search \
  -H "Authorization: Bearer $ARISELABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "How will Nvidia stock react to its earnings?",
    "k": 5,
    "as_of_date": "2026-08-20",
    "recency": "3d"
  }'

{
  "data": [
    {
      "url": "https://example.com/markets/nvidia-q2-preview",
      "title": "Options imply an 8% move for Nvidia into next week's report",
      "published_at": "2026-08-19T14:05:00Z",
      "text": "Options traders are pricing an 8% swing in either direction…"
    }
  ],
  "search_metadata": { "as_of_date": "2026-08-20", "returned_count": 5 }
}

Start with a key. Add a date when you need the past.

Start free