DeepSeek V4.1 Flash
Multimodal 1M-context MoE for coding agents
DeepSeek's V4.1 Flash: a 552B-backbone Mixture-of-Experts with 8–16B active parameters, native image understanding, a 1M-token context window, optional thinking mode, and native tool calling. It beats V4 Flash on every published agentic benchmark at a lower price.
Live playground
Try DeepSeek V4.1 Flash in your browser
Run the model, inspect its schema, then move the same request into production with one imageat API key.
Sign in to use the live model.
Your response will appear here
DeepSeek V4.1 Flash at a glance
DeepSeek's V4.1 Flash: a 552B-backbone Mixture-of-Experts with 8–16B active parameters, native image understanding, a 1M-token context window, optional thinking mode, and native tool calling. It beats V4 Flash on every published agentic benchmark at a lower price.
- Developer
- DeepSeek
- Released
- September 2026
- Architecture
- Mixture-of-Experts, causal encoder-decoder, compressed sparse attention 2
- Parameters
- 552B backbone, 8–16B active per token, plus 196B Engram memory
- Context window
- 1,000,000 tokens
- Input / output
- Text and up to 10 images in, text out
- Reasoning
- Thinking mode with low, medium, or high effort, or off
- Tool calling
- Native function calls, parallel calls
- Weights
- Open, MIT license
- imageat model ID
- deepseek/v4-1-flash
DeepSeek V4.1 Flash API pricing
API-key requests are billed on the tokens each request actually uses, rounded up to the cent. Reasoning tokens count as output, so thinking mode costs more than direct answers. In the playground, website credits are reserved and settle to the measured cost; most short answers use 1 credit.
Input
$0.60 / 1M tokens
Output (incl. reasoning)
$2.40 / 1M tokens
Need the deepest factual recall? Try DeepSeek V4 Pro.
See all API pricesV4 Flash vs V4.1 Flash
DeepSeek's published results for V4.1 Flash against V4 Flash. The biggest gains are in agent work: +20 points on DeepSWE, +17 on AutomationBench, and +12 on HLE with tools.
| Benchmark | Metric | V4 Flash | V4.1 Flash |
|---|---|---|---|
| GPQA Diamond | Pass@1 | 89.9 | 90.9 |
| Codeforces | Rating | 3289 | 3471 |
| MathArena Apex | Pass@1 | 58.6 | 65.6 |
| Terminal-Bench 2.1 | Pass@1 | 82.7 | 90.6 |
| DeepSWE v1.1 | Resolved | 54.4 | 74.2 |
| NL2Repo-Bench | Score | 54.2 | 64.0 |
| CyberGym | Pass@1 | 76.7 | 88.1 |
| HLE with tools | Pass@1 | 51.5 | 63.9 |
| AutomationBench | Pass@1 | 37.7 | 54.8 |
| Agent's Last Exam | Pass@1 | 25.2 | 31.8 |
Source: DeepSeek V4.1 Flash model card on Hugging Face. Scores are DeepSeek's own and were not reproduced by imageat.
Call the DeepSeek V4.1 Flash API
One OpenAI-compatible endpoint and your imageat API key. Point any OpenAI SDK, opencode, Cline, or the Vercel AI SDK at https://api.imageat.com/v1; the /v1 is required.
messagesSystem, user, assistant, and tool messages, OpenAI format.enable_thinkingfalse skips reasoning for the fastest, cheapest answer.reasoninglow, medium (default), or high thinking effort.max_tokensCaps the reply when thinking is off (up to 4,000).tools & tool_choiceOpenAI function definitions; replies with tool_calls.parallel_tool_callsAllow several tool calls in one turn (default true).streamOpenAI-style server-sent events.user & session_idOptional identifiers for your end users and sessions.
curl https://api.imageat.com/v1/chat/completions \
-H "Authorization: Bearer iat_live_YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek/v4-1-flash",
"messages": [
{"role": "system", "content": "You are a senior software engineer."},
{"role": "user", "content": "Find the bug in this function and fix it."}
],
"reasoning": "high"
}'from openai import OpenAI
client = OpenAI(base_url="https://api.imageat.com/v1", api_key="iat_live_YOUR_KEY")
response = client.chat.completions.create(
model="deepseek/v4-1-flash",
messages=[{"role": "user", "content": "Summarize this contract in five bullet points."}],
extra_body={"enable_thinking": False}, # fastest answer; max_tokens applies
max_tokens=800,
)
print(response.choices[0].message.content)DeepSeek V4.1 Flash questions
How much does the DeepSeek V4.1 Flash API cost on imageat?
+
API-key requests are billed at 2× DeepSeek's measured token cost, rounded up to the cent per request. During DeepSeek's peak hours (Mon–Fri 09:00–12:00 and 14:00–18:00 Beijing time) that is $0.60 per 1M input tokens and $2.40 per 1M output tokens; the rest of the time it is half. Reasoning tokens count as output. With website credits, most short answers settle to 1 credit.
What changed from DeepSeek V4 Flash?
+
V4.1 Flash reads images, has a larger 552B backbone with fewer active parameters per token (8–16B vs 13B), and scores higher on every agentic benchmark DeepSeek published, for example 74.2% vs 54.4% on DeepSWE v1.1. It also costs less per token.
Can DeepSeek V4.1 Flash read images?
+
Yes. Send OpenAI-format image_url content parts (JPEG, PNG, or WebP, up to 10 per request) next to your text. It reads screenshots, charts, and documents; it does not generate images.
What is the context window of DeepSeek V4.1 Flash?
+
1 million tokens. Your messages and the completion, including reasoning, share that window.
Is DeepSeek V4.1 Flash OpenAI-compatible?
+
Yes. Send model "deepseek/v4-1-flash" to https://api.imageat.com/v1/chat/completions with your imageat API key. OpenAI SDKs, opencode, Cline, and the Vercel AI SDK work by setting the base URL to https://api.imageat.com/v1, and streaming is supported.
How do I turn thinking off?
+
Send "enable_thinking": false for the fastest, cheapest answers; max_tokens then caps the reply. Leave it on and set "reasoning" to low, medium, or high to choose how long the model thinks first. Only the final answer is returned in message.content.
Does DeepSeek V4.1 Flash support tool calling?
+
Yes. Pass OpenAI-format tools and the response returns tool_calls with finish_reason "tool_calls", including several parallel calls in one turn.
What is DeepSeek V4.1 Flash Economy?
+
The same model on a slower, cheaper serving tier: 20% less per token, identical answers. Use model "deepseek/v4-1-flash-economy" for batch and background work where latency matters less.
Reason with context
Use the model for analysis, planning, and multi-step knowledge work.
Build and code
Apply it to software development, structured output, and agent workflows.
Generate text
Create clear answers, summaries, drafts, and conversational experiences.
What DeepSeek V4.1 Flash is built for
DeepSeek's V4.1 Flash: a 552B-backbone Mixture-of-Experts with 8–16B active parameters, native image understanding, a 1M-token context window, optional thinking mode, and native tool calling. It beats V4 Flash on every published agentic benchmark at a lower price.
- Coding agents and long tool-calling loops
- Screenshots, documents, and charts as image input
- Long-document Q&A across a 1M-token context
- Thinking on for hard problems, off for the fastest answers
Call DeepSeek V4.1 Flash from your app
Use the OpenAI-compatible chat endpoint with your imageat API key and pay-as-you-go balance.