DeepSeek V4 Pro vs V4.1 Flash: what is different and which to use

cURL

curl https://api.apimodels.app/v1/chat/completions \
  -H "Authorization: Bearer $APIMODELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4.1-flash",
    "messages": [{"role": "user", "content": "Explain this stack trace in two sentences."}]
  }'

Python

from openai import OpenAI

client = OpenAI(api_key="YOUR_APIMODELS_KEY", base_url="https://api.apimodels.app/v1")

# Run the same task on both and compare quality and token usage before you choose.
for model in ["deepseek-v4.1-flash", "deepseek-v4-pro"]:
    r = client.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": "Write a SQL query that returns the top 5 customers by revenue."}],
    )
    print(model, r.usage.total_tokens)
    print(r.choices[0].message.content)

What is the difference between DeepSeek V4 Pro and V4.1 Flash?

V4 Pro is the large V4 model (1.6T parameters, 49B active, GA on 13 August 2026). V4.1 Flash is newer (10 September 2026), a 552B mixture-of-experts model on a new architecture, and DeepSeek says it scores ahead of V4 Pro on benchmarks. Both have a 1M context, thinking mode and tool calls. On apimodels.app Flash costs $0.27 / $1.08 per 1M tokens at peak and Pro $1.247 / $3.742.

Is V4.1 Flash better than V4 Pro?

On DeepSeek’s own benchmarks, yes: its 10 September 2026 release note says V4.1 Flash delivers results ahead of flagship models including V4 Pro. Benchmarks are not your workload, so run your own prompts on both before moving production traffic; the switch is a one-string change.

How much more does V4 Pro cost than V4.1 Flash?

On apimodels.app, at peak, V4 Pro is $1.247 input and $3.742 output per 1M tokens against $0.27 and $1.08 for V4.1 Flash, roughly four times the input price and three and a half times the output price. Off-peak both halve: $0.624 / $1.871 for Pro and $0.135 / $0.54 for Flash.

When is off-peak, and is it the same for both?

Peak is 01:00–04:00 and 06:00–10:00 UTC (09:00–12:00 and 14:00–18:00 Beijing time). For V4.1 Flash those hours count as peak only from Monday to Friday, and weekends and Chinese public holidays are off-peak all day. For V4 Pro the same hours are peak on every day, weekends included.

Which model id should I use? Does deepseek-v4-flash still work?

Use deepseek-v4.1-flash or deepseek-v4-pro. deepseek-v4-flash still works and bills like V4.1 Flash, since DeepSeek retired V4 Flash and serves that name with V4.1 Flash, so existing code keeps running.