DeepSeek V4.1 Flash

Available

deepseek/deepseek-v4.1-flash

DeepSeek's fast, low-cost agentic coding model with a 1M-token context window.

Chat
1.05M context
131K max output
Tools

Try in Playground Compare

Pricing

Input: $0.15 / 1M tokens
Output: $0.60 / 1M tokens
Cached input: $0.005 / 1M tokens

What it costs

A typical request: 2,000 tokens in, 500 out$0.0006

1,000 of those requests$0.60

Performance on SparrowCodes

Measured from real requests over the last 7 days.

Throughput: 130 tok/s
First token: 1.16 s
Latency: 3.07 s
Success rate: 99.4%

Medians across 173 requests.

Official benchmarks

Scores published by the model's maker, under their test conditions.

The maker hasn't published benchmark scores we could verify for this model.

Specifications

Type: Chat
Served by: SparrowCodes
Context window: 1,048,576 tokens
Maximum output: 131,072 tokens
Input types: Text
Tool calling: Supported

Use it

curl https://api.sparrowcodes.com/v1/chat/completions \
  -H "Authorization: Bearer $SPARROW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4.1-flash",
    "messages": [{"role": "user", "content": "Explain rate limiting in one paragraph."}]
  }'

Set "model" to deepseek/deepseek-v4.1-flash. The response includes its exact cost.

More from DeepSeek

All models

From the blog

Meta Muse Ember 1.0: a $0.10-per-million-token model for agents, with a 1M context
Meta's smallest and cheapest Muse model, released 9 November 2026, is built for the routine parts of agent work: tool calls, routing, extraction and coding subtasks. Price, limits and how it compares.
Kimi K3.1 Pro is now open-weight: 2.8 trillion parameters, 1M context
Moonshot AI published the weights of Kimi K3.1 Pro on Hugging Face in early November 2026. It ranks first among open-weight models on Arena's agent leaderboard. Size, licence and what it takes to run.
Mistral Large 4 open weights: a 1-trillion-parameter model you can download
Mistral published the weights of Mistral Large 4 on 29 October 2026: about 1 trillion parameters, roughly 50 billion active, 1M context and image input. The licence, the API price and the hardware it needs.