Mistral Large 4

Mistral Large 4 (mistralai/mistral-large-4-0) is Mistral’s senior model and the brand’s first appearance in the Mixen catalog. It accepts images and supports the full sampling toolkit (temperature, top_p, seed, frequency_penalty, presence_penalty, logprobs) — most of which OpenAI and Claude models reject.

Parameters and capabilities

Parameter	Value
model	mistralai/mistral-large-4-0
Images on input	yes (image_url, data URLs)
Files on input	no
Reasoning	reasoning_effort: high; disabled with off
Cache read	$0.175 per 1M tokens
Sampling	temperature, top_p, seed, frequency_penalty, presence_penalty, repetition_penalty, logprobs

Example

Python

from openai import OpenAI

client = OpenAI(base_url="https://api.mixen.ai/v1", api_key="mxn-...")

resp = client.chat.completions.create(
    model="mistralai/mistral-large-4-0",
    messages=[{"role": "user", "content": "Write unit tests for this tax-id validation function"}],
    temperature=0.2,
    seed=42,
)
print(resp.choices[0].message.content)

cURL

curl https://api.mixen.ai/v1/chat/completions \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistralai/mistral-large-4-0",
    "messages": [{"role": "user", "content": "Write unit tests for this tax-id validation function"}],
    "temperature": 0.2,
    "seed": 42
  }'

Pricing

Per 1M tokens: input — $1.7, output — $5.225. Cache: read — $0.175. Showcase prices follow the live exchange rate — take fresh values from the catalog.

Strengths and limitations

• The full sampling toolkit: seed for reproducibility, logprobs for confidence analysis — things closed models do not accept.
• Single-level reasoning: only high or a full off — no middle depths.
• No file input: extract PDF/DOCX to text beforehand.

All models — in the catalog. The general workflow — chat guide.