gramm.ai

What It Is

Gramm AI is a load forecasting API for the U.S. power market, positioned as “auditable grid forecasting infrastructure.” It covers seven U.S. ISO/RTOs—CAISO, ERCOT, PJM, MISO, NYISO, ISONE, and SPP—and provides day-ahead forecasts plus 2/7/15-day forecasts where available. Outputs are MW load values, and accuracy metrics can also be returned.

Core Capabilities

Its AI capabilities are focused on grid demand forecasting rather than general-purpose generative AI. Documentation examples show the model as gramm-v22. Paid tiers can provide p10/p90 forecast intervals for some rows, and incorporate or track inputs such as weather, holidays, generation mix, and ISO-specific signals. The API design is relatively mature, with forecast, history, latest, accuracy, regions, and health endpoints. JSON is the default format, with CSV/XML and a CAISO OASIS-compatible interface also supported. The accuracy endpoint can return MAPE, RMSE, MAE, and bias, making it suitable for trading and risk teams that need auditing and backtesting.

Pricing and Trial

The free tier requires no credit card and has no expiration. It includes live API access, 100 requests/hour, 1,000 requests/day, and 2 API keys, but only supports day-ahead forecasts, hourly resolution, and no SLA. Developer is $199/month, Team is $1,999/month, Growth is $4,999/month, and Enterprise is custom-priced. The official site makes clear that there is no extra per-API-call billing; instead, it uses fixed subscriptions with different limits, making costs predictable.

Pros and Cons

Its strengths are a clearly defined vertical use case, complete API documentation, a low-friction free evaluation tier, support for live scorecards and accuracy metrics, and enterprise procurement features such as MSA, DPA, SSO, and dedicated support. The drawbacks are also clear: coverage is limited to seven U.S. grids; many capabilities are marked as “where available” or “validated,” so actual availability needs to be verified region by region; integrations with Molecule, OATI, AspectCTRM, and others are still in development or planning; and there is no visible Chinese interface, Chinese documentation, or support for China’s domestic power grids.

Who It’s For and Access from China

Gramm AI is best suited to U.S. power trading desks, REPs, IPPs, utilities, energy startups, mining operators, and quant teams for load forecasting, risk monitoring, bid optimization, OASIS migration, and regional benchmarking. Access from China is not disclosed in the available text, and payment methods are not specified. Judging by its business coverage, it is not aimed at the Chinese power market. If you only need research or a self-built baseline, alternatives or complementary options such as TimeGPT, EIA Open Data, statsforecast, or CAISO OASIS may be worth considering.