Requesty
Sign in

identity

open weights
Fireworks AI logo

deepseek-v4.1-flashFireworks AI

in /1M$0.22input tokens
out /1M$0.66output tokens
context1.0M393K out
median ttft1.12smeasured
api
chat
hosting
US
model lab
DeepSeek
weights
open
added
September 2026
model id
fireworks/deepseek-v4.1-flash

capabilities 5/8

visionreasoningtoolscachingweb searchjson schemacomputer useimage gen

3 providers serve deepseek-v4.1-flash. This page is one of them. Compare all endpoints

pricing

no markup
input
$0.22
output
$0.66
cache write
-
cache read
$0.0070

worked cost

Estimated cost for three request volumes
volumecost
100K in + 10K out$0.0286
1M in + 100K out$0.29
10M in + 1M out$2.86

Requesty charges what the upstream provider charges, with no markup and no per-request fee. Prompt caching and smart routing cut the effective cost further on repeated context. Gateway pricing

provider price, passed throughper 1M tokens

measured

through Sep 12
median ttft
1.12s
median tok/s
79.0
blended /1M
$0.03
cache hit
93.3%

Weekly aggregates from production traffic routed through Requesty to this endpoint. A metric with no qualifying sample is left out rather than filled in. Compare against the other providers

benchmarks

artificial analysis
Benchmark scores, each normalised to 100
evalScore as a share of 100/100
SciCode51.9
Intelligence Index39.5
Humanity's Last Exam39.2
3 evalsmean43.5

Scores from Artificial Analysis and public leaderboards, normalised to 100. Bar colour is the band, not the rank: green 80 and up, blue 55 and up, amber 30 and up. Benchmarks measure narrow skills, so test on your own workload before committing. Released 2026-09-10.

policy

data terms
retention
none
trains on prompts
no
hosted in
US

Terms are the provider's, not Requesty's: routing a request here puts it under them. Fireworks Privacy Policy

Need every request to stay in one jurisdiction? EU hosting routes only through EU-hosted providers.

quickstart

openai compatible
fireworks/deepseek-v4.1-flash

this id pins the fireworks ai deployment, with no routing and no failover. the managed id on the canonical page routes across all 3 providers instead

main.py
from openai import OpenAI client = OpenAI(    api_key="YOUR_REQUESTY_API_KEY",    base_url="https://router.requesty.ai/v1",) response = client.chat.completions.create(    model="fireworks/deepseek-v4.1-flash",    messages=[        {"role": "user", "content": "Explain quantum computing in one paragraph."},    ],) print(response.choices[0].message.content)

Change the base url to https://router.requesty.ai/v1, use your Requesty key, set the model to the id above. Existing OpenAI SDK code needs no other edit, and the same key reaches every other model in the catalog. Browse all models

notes

reference

DeepSeek V4.1 Flash is a 552B parameter Mixture of Experts model built on a new Causal Encoder Decoder architecture that activates only 8B parameters for input and 16B for output, giving frontier level intelligence at a fraction of the cost. It is the smallest model in the new DeepSeek architecture family and the first Flash with native vision and multimodal understanding. V4.1 Flash beats DeepSeek V4 Pro on performance, cost, speed and task completion time and scores 88.1 on CyberGym. Its KV cache is compressed to 890 bytes per token, four times smaller than V4 Flash and over 400 times smaller than V1, cutting the cache hit costs that dominate agent workloads. 1M token context window, 384K max output, thinking and non thinking modes, tool calling, structured JSON output and prompt caching. Served on Fireworks at $0.22 input, $0.007 cache read and $0.66 output per 1M tokens. Open weights are published on Hugging Face. Ideal for coding agents, long context RAG, high throughput batch processing and multimodal assistants.

questions 6

How much does deepseek-v4.1-flash cost?
deepseek-v4.1-flash is priced at $0.22 per million input tokens and $0.66 per million output tokens when accessed via Requesty. Prompt caching is supported, which can cut effective input cost by up to 90% on repeated context. Requesty charges exactly what the upstream provider charges, we don't add markup.
What is the context window of deepseek-v4.1-flash?
deepseek-v4.1-flash has a context window of 1.0M tokens, with a maximum output of 393K tokens per response. That is roughly 1,398 words of input you can fit in a single prompt.
How does deepseek-v4.1-flash perform on benchmarks?
deepseek-v4.1-flash scores 51.9% on SciCode, 39.5% on Intelligence Index, 39.2% on Humanity's Last Exam. The benchmarks pane shows every published result, each normalised to 100.
What can deepseek-v4.1-flash do?
deepseek-v4.1-flash supports vision input, tool calling, extended reasoning, prompt caching, structured outputs (JSON schema). You can call it through any OpenAI-compatible client by pointing base_url to Requesty.
How do I use deepseek-v4.1-flash with the OpenAI SDK?
Install the OpenAI SDK, set base_url to "https://router.requesty.ai/v1", set your API key to your Requesty key, and set the model to "fireworks/deepseek-v4.1-flash". The quickstart pane shows Python, JavaScript and cURL snippets.
Can I run deepseek-v4.1-flash through Requesty?
Yes. deepseek-v4.1-flash runs through Requesty's OpenAI-compatible API, served from Fireworks AI. You do not host the model yourself: point base_url at Requesty, set the model to "fireworks/deepseek-v4.1-flash", and requests are routed to the upstream provider with automatic failover. The same key gives you 600+ other models too.

more from fireworks ai 8

Other Fireworks AI endpoints
endpointctxin /M
deepseek-v4-flash-vision-exp1.0M$0.22
glm-5.3-flash1.0M$0.15
glm-5.31.0M$1.40
nemotron-lightning-3.5-30b-a3b262K$0.05
deepseek-v4-pro-08131M$1.32
muse-glimmer-30b131K$0.35
qwen3.8-max1M$2.00
deepseek-v4-flash-07311M$0.22

call deepseek-v4.1-flash through one endpoint

One key for this endpoint and 600+ other models. No markup on provider prices, automatic failover when a provider degrades, and prompt caching built in. Methodology

deepseek-v4.1-flashfireworks aiin $0.22 /1Mout $0.66 /1Mctx 1.0Mttft 1.12shosted US3 providers serve this model