Provider pricing hub

DeepSeek API pricing: V4 Flash, V4.1 Flash and V4 Pro.

DeepSeek API prices on RouterPlex run from $0.18/$0.73 per 1M input/output tokens (DeepSeek V4 Flash) to $0.80/$2.39 (V4 Pro), with 0% markup. V4 Flash, V4.1 Flash and V4 Pro each bill one flat rate, never the peak-hour double. Top-ups have a $0 fee, by card from $5 or crypto; balances stop at $0 and each key has a hard USD cap.

Prices checked 20 October 2026 from the live catalog. DeepSeek's pricing page last checked 20 Oct 2026: api-docs.deepseek.com/quick_start/pricing

DeepSeek on this key20 October 2026

Routes

$0.18

Cheapest input / 1M

Markup 0%

DeepSeek family

Published catalog rates.

Agent-turn cost uses 100K input and 10K output. Every rate here applies at every hour.

Model route	Context	Input / 1M	Cached input / 1M	Output / 1M	Agent turn
DeepSeek V4 Flash deepseek-v4-flash	1M	$0.18	$0.065	$0.73	$0.025
DeepSeek has retired V4 Flash: the id is now served by DeepSeek-V4.1-Flash and billed at the Flash price, $0.15 input / $0.60 output per 1M tokens off-peak and double that during peak hours (01:00-04:00 and 06:00-10:00 UTC, Monday to Friday). RouterPlex bills the same flat rate as deepseek-v4.1-flash, the time-weighted average of the two, so you are never billed by the clock.
DeepSeek V4.1 Flash deepseek-v4.1-flash	1M	$0.18	$0.065	$0.73	$0.025
DeepSeek prices this model by time of day: $0.15 input / $0.60 output per 1M tokens off-peak, and double that during peak hours (01:00-04:00 and 06:00-10:00 UTC, Monday to Friday). RouterPlex bills a single flat rate in between -- the time-weighted average of the two -- so you are never billed by the clock and never pay the peak rate for off-peak work.
DeepSeek V4 Flash Vision deepseek-v4-flash-vision-exp	1M	$0.22	$0.007	$0.66	$0.029
DeepSeek has retired this preview: it now serves the id with DeepSeek-V4.1-Flash and bills it at the Flash price ($0.15 input / $0.60 output per 1M tokens off-peak, double during peak hours). RouterPlex serves it through a separate upstream and bills a flat $0.22 input / $0.66 output per 1M tokens at all hours. For the same model at DeepSeek's time-weighted rate, use deepseek-v4.1-flash.
DeepSeek V4 Pro deepseek-v4-pro	1M	$0.80	$0.21	$2.39	$0.10
DeepSeek prices this model by time of day: $0.66 input / $1.98 output per 1M tokens off-peak, and double that during peak hours (01:00-04:00 and 06:00-10:00 UTC, Monday to Friday). RouterPlex bills a single flat rate of $0.80 / $2.39 in between -- the time-weighted average of the two -- so you are never billed by the clock. Cached prompt tokens are billed at $0.21 per 1M, set from what our upstream charges us rather than DeepSeek's cache-hit rate.

DeepSeek pricing calculator

Same workload. Every DeepSeek route.

Defaults are a 10K-in / 1K-out request, 3,000 times a month. Switch to the 100K / 10K agent turn if you are costing a coding loop.

Workload calculator

Catalog rates · USD · no markup

Input tokensOutput tokensRequests / month

Compare a 5.5% card top-up fee (OpenRouter's published card rate as of July 2026). Token rates stay the same; only the cash you put in changes.

Models in the estimate

3 selected

DeepSeek V4 Flash$0.18 / $0.73DeepSeek V4 Flash Vision$0.22 / $0.66DeepSeek V4 Pro$0.80 / $2.39DeepSeek V4.1 Flash$0.18 / $0.73

10K in + 1K out × 3,000 req$100 of tokens

Route	Per request	Monthly	$100 buys
DeepSeek V4 Flash DeepSeek	$0.0025	$7.59	39,525 req
DeepSeek V4.1 Flash DeepSeek	$0.0025	$7.59	39,525 req
DeepSeek V4 Pro DeepSeek	$0.010	$31.17	9,624 req

Estimates apply live catalog input and output rates only. Actual spend follows the tokens your client sends and the model returns. Image-generation routes are omitted because they are not a chat turn.

Peak vs off-peak

What DeepSeek charges by the clock, and what you pay here.

DeepSeek's peak window: 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday. Peak is double the off-peak rate. DeepSeek figures below are read from each route's catalog note.

Model route	DeepSeek off-peak / 1M	DeepSeek peak / 1M	RouterPlex / 1M	How it bills here
DeepSeek V4 Flash	$0.15 / $0.60	$0.30 / $1.20	$0.18 / $0.73	One flat rate, every hour
DeepSeek V4.1 Flash	$0.15 / $0.60	$0.30 / $1.20	$0.18 / $0.73	One flat rate, every hour
DeepSeek V4 Pro	$0.66 / $1.98	$1.32 / $3.96	$0.80 / $2.39	One flat rate, every hour

Cached input means prompt tokens DeepSeek has already seen in a recent request with the same prefix; they bill at the cached-input column in the main table instead of the full input rate.

Call it

Use DeepSeek from Claude Code or the OpenAI SDK.

Claude Code and the Anthropic SDK use https://api.routerplex.com. The OpenAI SDK, Cursor and Cline use https://api.routerplex.com/v1. Same key, same catalog id, same per-key budget. Swap deepseek-v4-flash for any id in the table.

Claude Code (shell)

export ANTHROPIC_BASE_URL=https://api.routerplex.com
export ANTHROPIC_AUTH_TOKEN="$ROUTERPLEX_API_KEY"
export ANTHROPIC_MODEL=deepseek-v4-flash
claude

OpenAI SDK (Python)

import os
from openai import OpenAI
 
client = OpenAI(
    base_url="https://api.routerplex.com/v1",
    api_key=os.environ["ROUTERPLEX_API_KEY"],
)
reply = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

How to pick

Which DeepSeek model for coding agents.

These notes only use catalog fields: price, context, and mode. They are not quality rankings.

Cheapest per agent turn

DeepSeek V4 Flash costs about $0.025 for a 100K-in / 10K-out turn. A long Claude Code session is mostly input, so the input and cached-input columns matter most.

No clock to watch

DeepSeek V4 Flash bills $0.18 / $0.73 per 1M at every hour. Off-peak at DeepSeek it is $0.15 / $0.60; peak is double. A job that runs into DeepSeek's peak window costs the same here.

Pro costs more per turn

DeepSeek V4 Pro is about 4.1x the agent-turn cost of DeepSeek V4 Flash. Same 1M context. Use it where your own evals show the difference.

Setup notes for DeepSeek V4 Flash Vision, Claude Code Router and the OpenAI-compatible API. Compare the rest of the catalog on all model pricing, the Kimi, Gemini, OpenAI and Anthropic hubs, the cheapest-first ledger, or RouterPlex vs OpenRouter.

Before you bill production

DeepSeek pricing, in questions.

01How much does the DeepSeek API cost?+

As of 20 October 2026, DeepSeek routes on RouterPlex cost $0.18 to $0.80 per 1M input tokens and $0.73 to $2.39 per 1M output tokens with 0% markup. DeepSeek V4 Flash is $0.18 / $0.73, DeepSeek V4.1 Flash is $0.18 / $0.73 and DeepSeek V4 Pro is $0.80 / $2.39. Top-ups have a $0 fee and the balance stops at $0.

02What are DeepSeek peak hours?+

DeepSeek prices some models by time of day. Its peak window is 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday, and the peak rate is double the off-peak rate. All other hours are off-peak.

03Will I pay the peak rate on RouterPlex?+

No. Every DeepSeek route here bills one catalog rate at every hour. DeepSeek V4 Flash, DeepSeek V4.1 Flash and DeepSeek V4 Pro are billed at a single time-weighted rate ($0.18 / $0.73, $0.18 / $0.73 and $0.80 / $2.39 per 1M) between DeepSeek's off-peak and peak prices, so you are never billed by the clock.

04What is DeepSeek cached input?+

Cached input is the part of a prompt DeepSeek has already processed in a recent request with the same prefix, such as a long system prompt or a repeated file. It bills at a lower rate than fresh input: $0.065 per 1M on DeepSeek V4 Flash, $0.065 per 1M on DeepSeek V4.1 Flash and $0.21 per 1M on DeepSeek V4 Pro in this catalog.

05Is DeepSeek V4 Pro worth it over V4.1 Flash?+

On price alone, DeepSeek V4 Pro costs about $0.10 per 100K-in / 10K-out agent turn against $0.025 for DeepSeek V4.1 Flash, about 4.1x. Both list 1M context. This page does not rank quality; compare answers on your own prompts before moving a coding loop to Pro.

06Can I use DeepSeek in Claude Code?+

Yes. Set ANTHROPIC_BASE_URL to https://api.routerplex.com, use your RouterPlex key as ANTHROPIC_AUTH_TOKEN, and set ANTHROPIC_MODEL to a catalog id such as "deepseek-v4-flash". The same key works with the OpenAI SDK at https://api.routerplex.com/v1. You do not need a DeepSeek account.

07Why might a number differ from DeepSeek's page?+

This page shows what RouterPlex bills, read from the live catalog on 20 October 2026. DeepSeek publishes separate peak and off-peak prices, renames and retires model versions, and changes rates on short notice. If the two disagree, the catalog row is what a request on your RouterPlex key costs.

Route DeepSeek from one balance

Try DeepSeek V4 Flash with $1, no card.

Try it with $1 of test credit, no card. Guided setup shows the exact Claude Code config for DeepSeek V4 Flash.

Eligible verified accounts get the $1 once. It never expires.

All routes Try it with $1, no card