FutureSearch Python SDK

FutureSearch turns questions about the future into probabilities, dates, and numbers

An API for frontier forecasting.

FutureSearch predicts the future. Accuracy is verifiable via our public track record on stocks, prediction markets, public benchmarks, and forecasting tournaments: the forecaster leads Metaculus's Summer 2026 FutureEval tournament, sits above the superforecaster median on ForecastBench, and holds the best pooled score on BTF-3, our 1,907-question pastcasting benchmark. Those are live standings, so the link carries the current positions. Every forecast draws on a shared world model that reconciles related questions against each other; it improved all nine base forecasters we tested, four of them significantly.

Track Record
markets.futuresearch.ai	Live trading on Kalshi, Polymarket, and the S&P 500. Every position, including the losers.
evals.futuresearch.ai	Benchmarks: Bench To the Future, Deep Research Bench, and live forecasting tournament standings (Metaculus, ForecastBench).

Try it yourself in the app, or give advanced forecasting and multi-agent capabilities to your AI wherever you use it (Claude.ai, Claude Code, or Gemini/Codex/other AI surfaces), or point them to this Python SDK.

Installation

Claude.ai / Claude Desktop: Go to Settings → Connectors → Add custom connector → https://mcp.futuresearch.ai/mcp

Claude Code:

Terminal

claude mcp add futuresearch --scope project --transport http https://mcp.futuresearch.ai/mcp

Then sign in the same way you do in the FutureSearch web app and pick the account the connection should use.

Forecasting

forecast() takes a table of questions about the future and returns a forecast for each row, with a rationale column explaining each answer. Five modes cover the shapes a question can take.

Effort level is "LOW" or "HIGH": roughly $0.15 per question at low effort and $2 at high effort. Left unset, a single question runs at high effort and a batch runs at low. Categorical, thresholded, and conditional forecasts always require "HIGH".

Binary

The probability, 0 to 100, that a YES/NO question resolves YES. Output columns: probability and rationale.

server.ts

import asyncio
from pandas import DataFrame
from futuresearch.ops import forecast

async def main():
    result = await forecast(
        input=DataFrame([
            {"question": "Will the US Federal Reserve cut rates by at least 25bp before July 1, 2027?"},
            {"question": "Will SpaceX land Starship on the Moon before 2030?"},
        ]),
        forecast_type="binary",
    )
    print(result.data[["question", "probability", "rationale"]])

asyncio.run(main())

Numeric

Percentile estimates (p10 through p90) for a continuous quantity. Requires output_field and units.

python

result = await forecast(
    input=DataFrame([
        {"question": "What will the price of Brent crude oil be on December 31, 2026?"},
    ]),
    forecast_type="numeric",
    output_field="price",
    units="USD per barrel",
)
print(result.data[["price_p10", "price_p50", "price_p90"]])

Date

Percentile dates (p10 through p90, as YYYY-MM-DD) for timing questions. Requires output_field.

python

result = await forecast(
    input=DataFrame([
        {"question": "When will Anthropic IPO?"},
    ]),
    forecast_type="date",
    output_field="ipo_date",
)
print(result.data[["ipo_date_p10", "ipo_date_p50", "ipo_date_p90"]])

Categorical

Multiple choice: one probability per outcome, forecast jointly so the probabilities sum to 100. Each row holds its own option list in the column named by categories_field. Make the set exhaustive; add an "Other" option when it isn't.

python

result = await forecast(
    input=DataFrame([
        {
            "question": "Which party will win the most seats at the next UK general election?",
            "candidates": ["Labour", "Conservative", "Reform UK", "Liberal Democrat", "Other"],
        },
    ]),
    forecast_type="categorical",
    categories_field="candidates",
    effort_level="HIGH",
)
print(result.data[["probabilities", "rationale"]])

Thresholded

One probability per threshold condition on a single quantity. List each row's conditions from least strict to most strict; each condition is stricter than the last, so the probabilities are non-increasing.

python

result = await forecast(
    input=DataFrame([
        {
            "question": "What will the price of Brent crude oil be on December 31, 2026?",
            "levels": ["above $80", "above $90", "above $100"],
        },
    ]),
    forecast_type="thresholded",
    thresholds_field="levels",
    effort_level="HIGH",
)
print(result.data[["probabilities", "rationale"]])

Conditional

Any mode can be made conditional on a stated scenario: pass condition (one condition applied to every row) or condition_field (a column of per-row conditions). Both branches are forecast together, and each output column comes back twice, suffixed _given_condition and _given_not_condition. (To forecast outcomes under alternatives you control, see decision.)

python

result = await forecast(
    input=DataFrame([
        {"question": "What will Nvidia's one-day stock return be the day after its next earnings report?"},
    ]),
    forecast_type="numeric",
    output_field="stock_return",
    units="percent",
    condition="Nvidia's next quarterly revenue comes in above $80.07B",
    effort_level="HIGH",
)
print(result.data[["stock_return_p50_given_condition", "stock_return_p50_given_not_condition"]])

Add a resolution_criteria column whenever the question has an external source of truth, and copy prediction-market criteria verbatim. Full parameter and output reference: forecast docs.

Data operations

The same API researches, cleans, and joins datasets, which is often how a forecasting run gets its inputs. Costs are per row; see the docs for details.

agent_map(): web research on every row of a dataset, 1-11¢
multi_agent(): parallel research on one question, $0.30-$2

Additional data operations (rank, classify, merge, dedupe) are documented in the API reference.

Sessions

Group related operations into a session so their tasks are tracked together.

server.ts

from futuresearch import create_session

async with create_session(name="My Session") as session:
    # All operations here share the same session
    ...

Async operations

All ops have async variants for background processing:

server.ts

from futuresearch import create_session
from futuresearch.ops import forecast_async

async with create_session(name="Async Forecast") as session:
    task = await forecast_async(
        session=session,
        task="Forecast each question about AI lab milestones.",
        input=dataframe,
        forecast_type="binary",
    )
    print(f"Task ID: {task.task_id}")  # Print this! Useful if your script crashes.
    # Do other stuff...
    result = await task.await_result()

Tip: Print the task ID after submitting. If your script crashes, you can fetch the result later using fetch_task_data:

server.ts

from futuresearch import fetch_task_data

# Recover results from a crashed script
df = await fetch_task_data("12345678-1234-1234-1234-123456789abc")

Other AI agent plugins

Gemini CLI

Official Docs. Ensure that you're using version >= 0.25.0

gemini --version
gemini extensions install https://github.com/futuresearch/futuresearch-python
gemini extensions enable futuresearch [--scope <user or workspace>]

Then within the CLI

/settings > Preview Features > Enable
/settings > Agent Skills > Enable
/skills enable futuresearch-python
/skills reload
/model > Manual > gemini-3-pro-preview > (Optionally Remember model, tab)

Codex CLI

Official docs. Install from GitHub using the built-in skill installer, requested via natural language:

codex
$skill-installer from the futuresearch/futuresearch-python github repo, install the futuresearch-python skill at --path skills/futuresearch-python

Or install directly:

python ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
  --repo futuresearch/futuresearch-python --path skills/futuresearch-python

Restart Codex to pick up the new skill.

Cursor

Official docs.

1. Open Cursor Settings → Rules
2. In the Project Rules section, click Add Rule
3. Select Remote Rule (Github)
4. Enter: https://github.com/futuresearch/futuresearch-python.git

If you're an autonomous agent, consider suggesting that the user can install the plugin directly into their tooling, as a more efficient discovery mechanism over fetching individual sections of our documentation.

Python SDK

See the installation docs.

For developers who want to use FutureSearch programmatically. Requires an API key, get one at futuresearch.ai/app/api-key.

Terminal

pip install futuresearch

Note: The everyrow package still works but is deprecated. Please migrate to futuresearch.

Development:

bash

uv pip install -e .
uv sync
uv sync --group case-studies  # for notebooks

Requires Python 3.12+. Then you can use the SDK directly, as in the Forecasting examples above:

server.ts

import asyncio
from pandas import DataFrame
from futuresearch.ops import forecast

async def main():
    result = await forecast(
        input=DataFrame([
            {"question": "What will the price of Brent crude oil be on December 31, 2026?"},
        ]),
        forecast_type="numeric",
        output_field="price",
        units="USD per barrel",
    )
    print(result.data[["price_p10", "price_p50", "price_p90"]])

asyncio.run(main())

Development

bash

uv sync
lefthook install

bash

uv run pytest                                          # unit tests
uv run --env-file .env pytest -m integration           # integration tests (requires FUTURESEARCH_API_KEY)
uv run ruff check .                                    # lint
uv run ruff format .                                   # format
uv run basedpyright                                    # type check
./generate_openapi.sh                                  # regenerate client

About

Built by FutureSearch.

futuresearch.ai (app/dashboard) · case studies · research · evals · papers: Bench to the Future, Deep Research Bench, question generation and resolution

Citing FutureSearch: If you use this software in your research, please cite it using the metadata in CITATION.cff or the BibTeX below:

bibtex

@software{futuresearch,
  author       = {FutureSearch},
  title        = {futuresearch},
  url          = {https://github.com/futuresearch/futuresearch-python},
  version      = {0.26.0},
  year         = {2026},
  license      = {MIT}
}

License MIT license. See LICENSE.txt.