impulse

DeepSeek V4.1 Flash

Open weights

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture.

API id DeepSeek-V4.1-FlashReleased 10 Sep 2026 (Sep 2026)
Databricks (except 1017) retires it on 19 Nov 2026 (tomorrow); Azure AI Foundry (except 1017) on 15 Dec 2026. 26 hosts keep serving it.
Input / 1M
$0.14
at DeepInfra
Output / 1M
$0.42
at DeepInfra
Context window
1M
tokens
Max output
131K
tokens per reply
Hosts
26
39 listings
Released
Sep 2026
10 Sep 2026

Pricing

List prices at DeepInfra, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.14$0.42
Cached input
97% off input
$0.0042

What it costs a month

At DeepInfra list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$61.28
  • RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$49.73
  • Coding agent1,000 requests a day, 40K in / 2K out, cached$79.13
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$437
Use your own numbers

Where to run it

26 hosts serve DeepSeek V4.1 Flash. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
SailCloud
DeepSeek-V4.1-Flash
$0.09 / $0.30
−32% vs headline
deepseek-v4.1-flash
$0.12 / $0.48
DeepInfraInference
DeepSeek-V4.1-Flash
$0.14 / $0.42
W&B InferenceInference
DeepSeek-V4.1-Flash
$0.13 / $0.65
+24% vs headline
SCX.aiCloud
DeepSeek-V4.1-flash
$0.15 / $0.60
+25% vs headline
PrismCloud
deepseek-v4.1-flash
$0.17 / $0.63
+36% vs headline
deepseek-v4.1-flash
$0.01 / $1.20
+46% vs headline
GMI CloudInference
DeepSeek-V4.1-Flash
$0.18 / $0.72
+50% vs headline
1017
deepseek-v4-1-flash-1017
$0.20 / $0.80
+67% vs headline
ai&Cloud
deepseek-v4.1-flash
$0.30 / $0.60
+79% vs headline
deepseek-v4.1-flash
$0.27 / $0.81
+93% vs headline
Mistral AILab API
1017
deepseek-v4-1-flash-1017
$0.30 / $1.20
+150% vs headline
deepseek-v4.1-flash
$0.30 / $1.20
+150% vs headline
Retires 15 Dec 2026
deepseek-v4.1-flash
$0.30 / $1.20
+150% vs headline
deepseek-v4.1-flash
$0.30 / $1.20
+150% vs headline
WallyCloud
deepseek-v4.1-flash
$0.30 / $1.20
+150% vs headline
ZeroGPUCloud
deepseek-v4.1-flash
$0.30 / $1.20
+150% vs headline
1017
deepseek-v4.1-flash-1017
$0.30 / $1.20
+150% vs headline
deepseek-v4.1-flash-maas
$0.30 / $1.20
+150% vs headline
BasetenInference
DeepSeek-V4.1-Flash
$0.30 / $1.20
+150% vs headline
DeepSeek-V4.1-Flash
$0.30 / $1.20
+150% vs headline
Together AIInference
DeepSeek-V4.1-Flash
$0.30 / $1.20
+150% vs headline
Fireworks AIInference
deepseek-v4p1-flash
$0.30 / $1.20
+150% vs headline
databricks-deepseek-v4-1-flash
$0.30 / $1.20
+150% vs headline
Retires 19 Nov 2026
DeepSeek-V4.1-Flash
$0.40 / $1.60
+233% vs headline
NVIDIA NIMInference
deepseek-v4.1-flash
-
Show 13 more listings: regions, snapshots and variants
DeepInfra
1017
DeepSeek-V4.1-Flash-1017
$0.14 / $0.42
W&B Inference
1017
DeepSeek-V4.1-Flash-1017
$0.30 / $1.20
+150% vs headline
GMI Cloud
1017
deepseek-v4.1-flash-1017
$0.18 / $0.72
+50% vs headline
ai&
1017
deepseek-v4.1-flash-1017
$0.30 / $0.60
+79% vs headline
Azure AI Foundry
1017
deepseek-v4.1-flash-1017
$0.30 / $1.20
+150% vs headline
Wally
1017
deepseek-v4.1-flash-1017
$0.375 / $1.50
+212% vs headline
Baseten
1017
DeepSeek-V4.1-Flash-1017
$0.30 / $1.20
+150% vs headline
Nebius AI Studio
1017
DeepSeek-V4.1-Flash-1017
$0.30 / $1.20
+150% vs headline
Together AI
1017
DeepSeek-V4.1-Flash-1017
$0.30 / $1.20
+150% vs headline
Fireworks AI
Alias
deepseek-v4p1-flash
$0.30 / $1.20
+150% vs headline
1017
deepseek-v4p1-flash-1017
$0.30 / $1.20
+150% vs headline
Databricks
1017
databricks-deepseek-v4-1-flash-1017
$0.30 / $1.20
+150% vs headline
Inference APIs
Alias
deepseek-v4.1-flash
$0.40 / $1.60
+233% vs headline

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
TextImages
Produces
Text
Features
Prompt caching
Open weights
License: DeepSeek License (MIT code)
Runs in about 1831.7 GB of GPU memory
Runtimes: SGLang, TGI, vLLM

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

  • Sail
    DeepSeek-V4.1-Flash
    $0.09 / $0.30
    32% less

Smaller DeepSeek models

No cheaper DeepSeek model keeps its key capabilities.

Other frontier labs

  • Claude Haiku 5.5
    Anthropic · on Anthropic
    $0.10 / $0.50
    Has every key capability
    5% less
  • GPT-6 Luna
    OpenAI · on OpenAI
    $0.10 / $0.50
    Has every key capability
    5% less

Open-weight picks

  • Qwen3.8 Flash
    Alibaba Qwen · on AIHubMix
    $0.113 / $0.38
    Has every key capability
    15% less
  • Nemotron 3.5 Super VL
    NVIDIA · on Kyma API
    $0.081 / $0.472
    Has every key capability
    15% less
  • MiMo V2.6 Flash
    Xiaomi · on GMI Cloud
    $0.112 / $0.224
    Has every key capability
    33% less
  • MiniMax M3.1 Flash
    MiniMax · on Refuseless
    $0.05 / $0.25
    Has every key capability
    52% less

Model line

Newer and older versions, at each lab's own price.

  1. Sep 2026DeepSeek V4.1 FlashLatest$0.14 / $0.42
  2. Apr 2026DeepSeek V4 Flash$0.30 / $1.20

Call it

Using DeepInfra's API.

Python
from litellm import completion

response = completion(
    model="deepinfra/DeepSeek-V4.1-Flash",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does DeepSeek V4.1 Flash cost?

On DeepInfra, DeepSeek V4.1 Flash costs $0.14 per million input tokens and $0.42 per million output tokens. Cached input costs $0.0042 per million tokens. The cheapest host is Sail, at $0.09 per million input tokens and $0.30 per million output tokens.

Which APIs offer DeepSeek V4.1 Flash?

DeepSeek V4.1 Flash runs on 26 APIs: Sail, CLF AI Gateway, DeepInfra, W&B Inference, SCX.ai, Prism, CoralBricks, GMI Cloud, EmpirioLabs AI and 17 more.

What is DeepSeek V4.1 Flash's context window?

1M tokens, with up to 131K tokens in one reply.

Is DeepSeek V4.1 Flash being retired?

Databricks (except 1017) retires it on 19 Nov 2026; Azure AI Foundry (except 1017) on 15 Dec 2026. 26 hosts keep serving it.

Is there a newer version of DeepSeek V4.1 Flash?

No. DeepSeek V4.1 Flash is the newest model in its line.

What are cheaper alternatives to DeepSeek V4.1 Flash?

Models that cost less and keep most of what it can do: Claude Haiku 5.5 on Anthropic (5% less), GPT-6 Luna on OpenAI (5% less) and Qwen3.8 Flash on AIHubMix (15% less). DeepSeek V4.1 Flash itself costs 32% less on Sail. Savings are on a mix of 3 input tokens to 1 output token.

When was DeepSeek V4.1 Flash released?

DeepSeek released it on 10 Sep 2026.

What is DeepSeek V4.1 Flash's API model id?

Use DeepSeek-V4.1-Flash with DeepInfra. Other hosts use their own ids: deepseek-v4.1-flash on CLF AI Gateway, DeepSeek-V4.1-flash on SCX.ai, deepseek-v4.1-flash on Prism, deepseek-v4.1-flash on CoralBricks.

Sources

Raw catalog entry for DeepSeek-V4.1-Flash on DeepInfra
{
  "max_input_tokens": 1048576,
  "max_output_tokens": 131072,
  "input_cost_per_token": 1.4e-7,
  "output_cost_per_token": 4.2e-7,
  "cache_read_input_token_cost": 4.2e-9,
  "supports_prompt_caching": true,
  "litellm_provider": "deepinfra",
  "mode": "chat",
  "supports_function_calling": true,
  "supports_tool_choice": true,
  "supports_response_schema": true,
  "supports_reasoning": true,
  "supports_vision": true,
  "source": "https://deepinfra.com/deepseek-ai/DeepSeek-V4.1-Flash"
}