About Mistral Small 4
Mistral Small 4 collapses what were three Mistral product lines into one open release: its model card describes a hybrid that acts as instruction model, reasoning model and coding specialist in a single Apache 2.0-licensed system, with a reasoning-effort control to move between modes. Architecturally it is a mixture-of-experts with a notably small active set per token drawn from a large expert pool, plus image input and strong system-prompt adherence.
It succeeds the earlier compact Mistral and Magistral models as the company's recommended small tier for chat assistants, coding and agentic tasks.
On the API, Small 4 is priced as a true budget tier — rates comparable with the cheapest hosted models anywhere — while the unified design means that price now includes reasoning capability that previously required a separate Magistral deployment. The reasoning-effort control is the cost lever to learn: deliberation bills as output tokens, so effort settings translate directly into money, and most routine traffic runs happily at the low end. Mistral's derived cache discount applies. Between the open weights, the low list rates and the built-in reasoning, it competes as much with self-hosting a mid-size model as with rival APIs.
Frequently asked questions
How much does the Mistral Small 4 API cost?
Mistral Small 4 costs $0.15 per million input tokens and $0.6 per million output tokens, with cached input at $0.015 per million (90% off). Note: Cached input derived from Mistral’s stated −90% discount.
How much does a typical chat message cost with Mistral Small 4?
A short chat message (about 500 input and 250 output tokens) costs roughly $0.0002 with Mistral Small 4. At 1,000 such requests a day, that is about $6.75 per month.
Is Mistral Small 4 cheap compared to similar models?
Mistral Small 4 has the #3 cheapest input rate of the 29 models we track. On a mixed workload (10K input / 2K output), it costs $0.0027 per request, versus $0.0013 for GPT-5 nano.
What does Mistral Small 4 cost per 1,000 tokens?
Per 1,000 tokens, Mistral Small 4 costs $0.00015 for input and $0.00060 for output, or $0.00001 for cached input. Providers quote rates per million tokens ($0.15/M in, $0.6/M out here), so divide by a thousand for the per-1K figure older pricing pages used.
How large a prompt can Mistral Small 4 take?
Mistral Small 4 accepts up to 256K tokens of input context. At its input rate, filling the entire 256K-token window costs about $0.039 per request before any output.
Related
Compare all models at once ·Count tokens in your prompt · Run Mistral Small 4 locally — VRAM requirements
All Mistral model costs: Mistral Medium 3.5 · Mistral Large 3
Last updated 2026-09-03. Prices verified against Mistral's official pricing page; see the methodology.