WORKCAST · Forecast board

AI demand,
forecast before it happens.

Every week we publish a dated, tamper-proof forecast of how much AI the leading apps will use over the next eight weeks, then score it in public against the standard forecasts and against GPT-6 Astra on its own. Our forecast combines the WORKCAST simulation with GPT-6 Astra and a statistical trend model. Misses stay up.

2 Nov
first forecast publishes
9 Nov
first live score lands
21%
less error one week ahead in the backtest (1,600 forecasts)

This week's call

First forecast publishes
Monday 2 November
as soon as OpenRouter publishes the week, and never more than two days after its cutoff.

Track record

Held-out test: the combined method

Weights fixed on earlier weekly cutoffs, then scored once on 7 new weekly cutoffs (2026-09-07 to 2026-10-19), all after GPT-6 Astra's knowledge cutoff: the lowest mean error of every method (23.9%). At 1 and 2 weeks ahead the damped trend was marginally ahead. Eight-week results come from the live board.

WORKCAST combinedGPT-6 Astra aloneBest standard forecast
1 week ahead7 held-out cutoffs · best other: damped trend
23.0% error
23.7%
23.0%
2 weeks ahead6 held-out cutoffs · best other: damped trend
25.6% error
26.5%
25.4%
4 weeks ahead4 held-out cutoffs · best other: damped trend
23.0% error
24.0%
23.9%
How it works

At each Monday cutoff, WORKCAST carries every leading app forward 8 weeks from data before the cutoff: its growth path, whether it stays among the leaders, and the models its workloads move to as new models launch and prices change. The point forecast is the simulation's central path; the range is the 5th to 95th percentile of 20 runs. Standard forecasts are made from the same data: the last four weeks carried forward, each app's recent trend, and market growth. Error is token-weighted across apps (lower is better).

How to verify a forecast

Each forecast is a read-only JSON record carrying the code version and a hash of every input table. Its SHA-256 is anchored in the Bitcoin blockchain with OpenTimestamps before the weeks it predicts. Download a record and its .ots proof, then run ots verify. A forecast published more than two days after its cutoff is skipped, never backdated.