Claude FM: Claude Forecasting Model

Rustam JamilovEconomics Fellow

Peter McCroryAnthropic (Mentor)

October 5, 2026

Claude's Real-Time Forecasts of the US Economy

Outcome, first release SPF median Claude Claude 80 percent interval

Forecast accuracy summary statistics

MAE: Mean absolute error, relative to first published outcomes. Lower is better.
RMSE: Root mean squared error, relative to first published outcomes. Lower is better.
Claude better: The share of forecasts where Claude's MAE is smaller than the SPF's.

Live US Forecasting Track: Committed until 2027Q3

Outcome, first release SPF median Claude

Model Cutoff Dates

Claude model	Training cutoff	Obs.	Blind	Blind, with outcome
Haiku 4.5	31 July 2025	60	60	18
Sonnet 5	31 January 2026	60	60	18
Fable 5.0	31 January 2026	60	60	18
Opus 5	31 May 2026	60	20	2
Sonnet 5.5	30 June 2026	60	20	2
Opus 5.5	30 June 2026	60	20	2

Notes: Each model forecasts 60 observations over the 3 rounds: four variables at five horizons in each round. A forecast is blind when the round's information date follows the model's training cutoff, so that nothing the model was trained on postdates the data it was given. Opus 5, Sonnet 5.5 and Opus 5.5 have the fewest because their cutoffs are the latest: the 2026Q1 and 2026Q2 rounds predate them.

Live track: forecasts against the published outcomes

Variable	Target	Obs.	Outcome	Claude error	SPF error	Winner
CPI inflation	2026Q1	3	3.19	0.369	0.541	Claude
CPI inflation	2026Q2	6	6.07	3.283	1.735	SPF
Real GDP growth	2026Q1	3	1.99	0.085	0.660	Claude
Real GDP growth	2026Q2	6	1.50	0.519	0.601	Claude
Unemployment rate	2026Q1	3	4.33	0.006	0.067	Claude
Unemployment rate	2026Q2	6	4.27	0.106	0.153	Claude
Unemployment rate	2026Q3	12	4.13	0.165	0.206	Claude
3-month Treasury bill	2026Q1	3	3.59	0.008	0.057	Claude
3-month Treasury bill	2026Q2	6	3.62	0.109	0.045	SPF
3-month Treasury bill	2026Q3	12	3.80	0.254	0.184	SPF
All pooled		60		0.509	0.398	SPF
By model, the same observations
Haiku 4.5		18		0.595	0.435	SPF
Sonnet 5		18		0.552	0.435	SPF
Fable 5.0		18		0.534	0.435	SPF
Opus 5		2		0.038	0.067	Claude
Sonnet 5.5		2		0.050	0.067	Claude
Opus 5.5		2		0.042	0.067	Claude

Notes: Forecasts with a published outcome, from the 2026Q1, 2026Q2 and 2026Q3 rounds. Each observation is one model's forecast of the target from one round, the three draws averaged. A forecast counts only when the round's information date follows the model's training cutoff. Errors are mean absolute errors against the first published estimate. Claude is closer than the SPF median on 7 of the 10 targets, and pooled over the 60 observations its error is 0.509 against the SPF's 0.398. Outcome convention: 2025Q4 averaged over its 2 published months, as FRED aggregates it; October 2025 was never collected.