Mistral has released Timestral, its first family of time-series foundation models, in four sizes from 5M to 209M parameters under a Modified MIT licence with a public GitHub repository, and states that the 209M model ranks first among pretrained, zero-shot models on GIFT-Eval, TIME and fev-bench.

● OPEN · settles 2026-10-21

Background

Mistral has stepped outside language models. Timestral is its first family of time-series foundation models: four sizes, at 5 million, 28 million, 123 million and 209 million parameters, all sharing one architecture and training recipe, built to forecast numbers rather than text. The models produce probabilistic forecasts across 21 quantiles, work zero-shot on series they were never trained on, and take in related series jointly, including drivers known in advance, such as production schedules or weather forecasts. Mistral says the 209M model ranks first among pretrained, zero-shot models on the field's three main public benchmarks — GIFT-Eval, TIME and fev-bench — across all five reported metrics, from a single checkpoint with no ensembling, at about two-thirds of the parameters of the previous best; the results have been submitted to the GIFT-Eval and TIME leaderboards. None of the four needs a GPU to run: on an eight-thread server CPU, a one-day-ahead forecast from three weeks of quarter-hourly readings takes 140 milliseconds with the largest model and 10 milliseconds with the smallest. Most of the training data was synthetic, generated first as single series with trend, seasonality and noise, then combined into multivariate sets with nonlinear dependencies, before a second stage on a mix of real and synthetic data at four times the context length, which reaches 32,768 time steps. The models are released under a Modified MIT licence, a public GitHub repository lets anyone run them on their own data, and a technical report is promised but not yet out. Organisations that want the models adapted to their own data are pointed to Mistral Forge, with deployment on-premises, in a VPC or air-gapped.

This market exists because launch announcements are easy to make and hard to grade. Press releases describe what a company intends to ship; they rarely come back later to say whether it shipped. AI Rank Bank pulls one checkable sentence out of the coverage — the sentence above — gives it a date, and lets readers price it in public. The launch itself stays with its publisher; what lives here is the claim, the crowd's call, and eventually the receipt.

How this claim is judged

Judging criteria, fixed when the market opened: Delivered if, at the check time, the Timestral models and GitHub repository remain publicly available under the Modified MIT licence in the four announced sizes, and Mistral's post still reports the 209M model ranking first among pretrained, zero-shot models on GIFT-Eval, TIME and fev-bench, with results submitted to the GIFT-Eval and TIME leaderboards. The release having been withdrawn, or the ranking claim removed or contradicted by the leaderboards as submitted, does not count.

The source of record is the publisher's own page linked above. A claim about a company's product is judged against that company's public output — its product pages, documentation, changelogs and official posts — not against rumors or third-party summaries. Only yes-or-no outcomes count: partial credit is not a verdict, and a claim that needs a paragraph of excuses has already answered itself.

How readers called it

1 reader called this claim. 0% said it would deliver.

0 Will deliver · 1 Won't ship

Points reward being right against the crowd. A correct call pays 100 × (1 − the share of readers who agreed with you when you placed it), with a floor of 10: side with a 90% consensus and a win pays 10, stand alone and a win pays up to 100. Wrong calls pay nothing. The price is locked in when the call is placed, so early readers are paid for judging before the crowd forms, not for copying it.

How readers split

The split line starts with this market's first snapshots — a point is recorded whenever a reader calls the claim, and at least once an hour while it stays open. Check back after the crowd has moved and the line will be here.

Verdict

No verdict yet. This claim is checked on 2026-10-21 08:00 Eastern Time (12:00 UTC) against the source of record below. When the check happens, one of three stamps lands on this page permanently: delivered if the claim came true as stated, busted if it did not, or unproven if the date passes without public evidence either way. Unproven claims pay no points and do not count against any reader's hit rate.

About this market

AI Rank Bank is an AI news front page with a memory. Every launch on the homepage carries one claim like this one, with a check date. Readers call claims with points, build a public record, and climb the leaderboard. Companies build a record too: every settled claim lands on the company's bank page, where its delivery rate is simple arithmetic that nobody gets to edit afterwards. There is no money in the market — points buy bragging rights only, which keeps the incentives pointed at being right.