Flash — fastest and cheapest, lighter reasoningOpen Weights

Mistral Small 4 (batch) Pricing

Mistral AI · mistralai-mistral-small-2603-batch

ShareX Facebook LinkedIn

Input / 1M tokens

$0.075

Output / 1M tokens

$0.300

Context window

262,144 tokens

≈ 350 pages of text

Max output

209,715 tokens

What it costs in practice

Typical request (1,200 in + 400 out tokens)$0.0002

1,000 requests / month$0.21/mo 10,000 requests / month$2.10/mo 100,000 requests / month$21.00/mo

Estimate your own workloadOpens the calculator with Mistral Small 4 (batch) preloaded — adjust volume and token counts there.

Price history

List price per 1M tokens since we started tracking in Aug 2026. Each step marks a price change — hover a point for the exact date and rate.

What it excels at

Cost-effective for high-volume batch text and vision workloads with very large context windows. Open weights enable self-hosted deployment at scale.

The business tradeoff

Smaller parameter count limits performance on complex multi-step reasoning relative to Mistral's larger models. Output quality and latency vary with hosting infrastructure.

Cheaper from Mistral AI

Mistral Small 3.2 24BMistral AI$0.0002/ typical request Ministral 3 8B 2512 (batch)Mistral AI$0.0001/ typical request

Same tier from other providers

GPT-5.4 NanoOpenAI$0.0007/ typical request Claude Haiku 4.5Anthropic$0.0032/ typical request Gemini 3.5 FlashGoogle$0.0054/ typical request

Head-to-head

Mistral Small 4 (batch) vs GPT-5.4 Nano →Mistral Small 4 (batch) vs Claude Haiku 4.5 →Mistral Small 4 (batch) vs Gemini 3.5 Flash →

Vendor list rates, as of Sep 10, 2026 · source: openrouter · per-request examples assume 1,200 input + 400 output tokens.