Model board / Mistral Small 4

Mistral Small 4

latest Mistral

Verified 2026-09-03 against the pages listed below. Prices are USD per 1M tokens, list rate — no committed-use, enterprise or volume discount.

Input
$0.15
Cached input
$0.015
Cache write
Output
$0.6
Context window
Max output
Weights
open
Caching
unverified

cached input $0.015 per 1M, printed per model on docs.mistral.ai/inference/pricing — the third Mistral pricing page, added to this row's sources this pass. It is the −90% discount the other two pages describe only in prose

not stated on the sourced pricing page

a per-model cached-input rate is published, but no page this row cites says whether caching is automatic or explicit, states a TTL, or names a minimum prefix — so the mechanics stay unverified even though the rate no longer is

Licence: Apache-2.0 (v26.03)

Watch out

Hybrid instruct/reasoning/coding model — pin the mode or output length varies

What we would use it for

Our most-reached-for fine-tuning base — cheap, small, open lineage

What this has cost, week by week

Output price at each refresh since 2026-07-30. A vendor change is the provider changing its price. Our correction means this board changed what it prints and the provider's charge did not move — the two are never merged, because publishing the second as the first would be a price rise that never happened.

07-30 $0.608-06 $0.608-14 $0.608-20 $0.608-27 $0.609-03 $0.6
our correction2026-08-06 · weights — As above — Apache-2.0 (v26.03) confirmed against the models doc.
newly stated2026-09-03 · cached_in — $0.015 from the same newly cited page. No change to what Mistral charges.

Where these numbers came from

No number on this page was written that was not read from one of those pages on 2026-09-03. If one is wrong, send us the source URL and it is corrected in the next weekly refresh, with a changelog line saying what was wrong — contact@naderu.com.

Jobs this is one of our picks for

Discussed in Naderu Weekly

The rest of the board

This row sits alongside 53 others on the model board, refreshed every week. The same data as JSON, and the terms for using it, are at /board-data.