DeepSeek V4.1 Flash is now available on AIHubMix and OpenRouter. It supports text and image inputs, a 1M-token context window, tool calling, structured outputs, and agent-oriented workflows.
AIHubMix currently offers a limited-time 30% discount on selected provider routes, valid through September 27, 2026. OpenRouter lists the price at $0.13 per million input tokens and $0.52 per million output tokens, with cache reads at $0.0026 per million tokens.
DeepSeek V4.1 Flash API pricing at a glance
| Platform / route | Input price | Output price | Cache read |
|---|---|---|---|
| OpenRouter | $0.13 / 1M | $0.52 / 1M | $0.0026 / 1M |
| AIHubMix — Bytedance off-peak | $0.0986 / 1M | $0.3942 / 1M | $0.002 / 1M |
AIHubMix’s Bytedance off-peak price reflects the current 30% discount, reduced from $0.1408 input and $0.5632 output per 1M tokens. The Bytedance off-peak schedule is 13:00–15:00 and 19:00–10:00 GMT+9.
What does the 30% discount save?
For a workload using 10 million input tokens and 2 million output tokens, the Bytedance off-peak discounted route costs approximately $1.7756. The corresponding peak route costs approximately $3.548. Before the discount, the same off-peak usage would cost $2.9764.
Does DeepSeek V4.1 Flash support long context?
Yes. DeepSeek V4.1 Flash supports a context window of approximately 1,048,576 tokens. This is useful for large codebases, long technical documents, multi-step research, persistent agent conversations, browser workflows, and large collections of logs or business data.
What makes DeepSeek V4.1 Flash different?
DeepSeek V4.1 Flash is a sparse mixture-of-experts model built on DeepSeek’s Causal Encoder-Decoder architecture. OpenRouter describes a 552B-parameter backbone, with approximately 8B parameters activated for input processing and 16B activated for output generation.
The model also includes native visual understanding. Text and image representations are trained jointly, allowing the model to process images alongside text.
Multimodal input and tool calling
DeepSeek V4.1 Flash accepts text and image input and returns text output. It supports function calling through tools and tool_choice, as well as structured outputs using JSON Schema through response_format.
This makes it suitable for coding agents, terminal agents, browser and computer-use systems, structured data extraction, visual document analysis, automated research, and multi-step business process automation.
AIHubMix vs OpenRouter for DeepSeek V4.1 Flash
AIHubMix
AIHubMix provides an OpenAI-compatible API, a unified model ID, multiple provider routes, time-based peak and off-peak pricing, provider monitoring, automatic retry and fallback, streaming, tool calling, structured outputs, and multimodal input.
OpenRouter
OpenRouter provides the model ID deepseek/deepseek-v4.1-flash, multi-provider routing, provider selection and exclusion, automatic failover, and public provider-level pricing and performance data.
OpenRouter currently lists multiple providers for the model, including Relace, Morph, DeepInfra, DeepSeek, Sail Research, Wafer, Fireworks, and StreamLake.
How to use DeepSeek V4.1 Flash through AIHubMix
AIHubMix provides an OpenAI-compatible endpoint:
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["AIHUBMIX_API_KEY"],
base_url="https://aihubmix.com/v1",
)
response = client.chat.completions.create(
model="deepseek-v4.1-flash",
messages=[
{"role": "user", "content": "Analyze this codebase and identify the highest-priority reliability risks."}
],
max_tokens=1024,
stream=False,
)
print(response.choices[0].message.content)
The AIHubMix model ID is deepseek-v4.1-flash. The OpenRouter model ID is deepseek/deepseek-v4.1-flash.
What is DeepSeek V4.1 Flash best used for?
DeepSeek V4.1 Flash is suitable for coding assistants, repository-level code analysis, terminal and computer-use agents, document and chart understanding, image-aware customer support, structured extraction, research agents, multi-step automation, and long-running workflows with repeated tool calls.
Frequently asked questions
Is DeepSeek V4.1 Flash free?
No. It is a paid API model. The AIHubMix promotion reduces the price on eligible routes but does not make the model free.
Does it support images and function calling?
Yes. It accepts text and image inputs, supports tools and tool_choice, and can return structured JSON outputs.
When does the AIHubMix discount end?
The AIHubMix model page currently states that the 30% promotion ends on September 27, 2026. Verify the model page before committing to a long-running workload.
The practical takeaway
DeepSeek V4.1 Flash combines a 1M-token context window, native image understanding, tool calling, structured outputs, and agent-oriented capabilities at a low per-token cost.
For AIHubMix users, use deepseek-v4.1-flash and check the provider’s GMT+9 schedule before estimating costs.
