Is DeepSeek V4 Pro Still Available? API Continues After Sep 14

Introduction

Search Is DeepSeek V4 Pro still available, and most people are worried about one line from the September 10 announcement: starting September 14, every deepseek-v4-pro request would route to the new Flash and bill at Flash rates.

Short answer: yes, it is still available. DeepSeek later noted on the Change Log and pricing page that, in response to demand, API service for DeepSeek V4 Pro continues after September 14, 2026, with billing unchanged; further changes will be announced. This site is an independent community—treat official pages as source of truth. As of writing (2026-09-17), the pricing table still lists both deepseek-flash and deepseek-v4-pro.

I. Timeline: “Will Route” First, Then “Continues”

Date (2026)	Public fact	What it means for you
Sep 10	DeepSeek-V4.1-Flash GA; API model name becomes deepseek-flash	Use the new ID for the daily tier
Same-day launch post	Planned: from 04:00 UTC Sep 14, deepseek-v4-pro would route to V4.1 Flash at Flash prices	Sparked “is Pro gone?” searches
Follow-up footnote	Change Log / pricing footnote: V4 Pro API continues after Sep 14; billing unchanged	Trust the footnote, not only the old routing sentence
After Sep 14	As of writing, pricing still lists deepseek-v4-pro (DeepSeek-V4-Pro-0813)	Existing integrations can keep the same model string

In short: lifecycle policy comes from the pricing footnote and Change Log, not a single unedited launch sentence. Prior picker guide: What is DeepSeek V4 Pro. The beta temp ID is covered in V4.1 Flash beta—GA daily tier is now deepseek-flash; do not ship expires-on-0910 in production.

II. How to Call It Now: Two Formal Model Names

base_url is unchanged (OpenAI-compatible https://api.deepseek.com). Use only names still on the docs table:

Model name	Backend version (as of writing)	Typical use
deepseek-v4-pro	DeepSeek-V4-Pro-0813	Flagship reasoning, long agent loops; lower concurrency cap
deepseek-flash	DeepSeek-V4.1-Flash	Daily throughput, native multimodal vision; legacy deepseek-v4-flash / deepseek-v4-flash-vision-exp temporarily route here

Minimal call (Pro still live):

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_KEY",
    base_url="https://api.deepseek.com",
)

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "Confirm in three sentences whether you are still V4 Pro."}],
)

print(response.choices[0].message.content)

Keys only from the official platform. Overview: DeepSeek API. Thinking toggle: DeepSeek thinking mode.

III. Pro vs Flash: Do Not Chase “Who Is Stronger” Alone

The launch post said third-party tests put V4.1 Flash ahead of V4 Pro on performance, cost, speed, and total runtime—useful for picking, not proof that the Pro endpoint is dead. Current USD per million tokens (peak/off-peak; verify on the official page):

Tier	Input · cache hit (off / peak)	Input · miss (off / peak)	Output (off / peak)	Concurrency (docs)
deepseek-flash	0.003 / 0.006	0.15 / 0.30	0.60 / 1.20	2500
deepseek-v4-pro	0.022 / 0.044	0.66 / 1.32	1.98 / 3.96	500

Practical split:

• Keep a locked Pro integration / contract flagship tier: stay on deepseek-v4-pro; service and billing continue as stated.
• Want lower unit price, higher concurrency, native vision: move to deepseek-flash, then stack off-peak and cache.
• Shift flexible jobs off-peak: peak is Mon–Fri UTC 01:00–04:00 and 06:00–10:00 (about 09:00–12:00 and 14:00–18:00 Beijing). Off-peak is half price—see API price change and context caching.

IV. Developer Checklist After September 14

1. Confirm logs still show deepseek-v4-pro. If something silently became Flash, check your own gateway map, then whether the official pricing table still lists Pro.
2. Confirm invoices still bill the Pro tier. “Billing unchanged” means Pro stays on Pro rates—not Flash rates under a Pro name.
3. Do not expect retired Flash behavior. Legacy deepseek-v4-flash still works but serves V4.1 Flash; new code should use deepseek-flash.
4. Rate limits follow the tier. Pro concurrency is lower—see API 429 rate limit.
5. Watch further notices. Official text says changes will be announced—until V4.1 Pro ships, trust the Change Log over “already retired” rumors.

V. FAQ

The launch post still says “will route to Flash”—which wins?
The pricing-page footnote and Change Log “continues to provide” paragraph. If the news post was not edited, the footnote is the effective policy.

Is DeepSeek V4 Pro still available on the web app too?
This article is about API model service. App/Web model switchers follow official product UI; site entry: DeepSeek web.

Could policy flip again in a few days?
No new end date was published. Keep model names swappable; watch the official Change Log. “Still available today” is not a forever guarantee.

Summary

Is DeepSeek V4 Pro still available? Yes: after Sep 14 the API continues, deepseek-v4-pro billing unchanged; prefer deepseek-flash for daily throughput and vision. Open the pricing table, confirm both rows, then decide whether to migrate the model string.

Further reading

• What is DeepSeek V4 Pro
• DeepSeek V4.1 Flash beta
• DeepSeek API price change
• DeepSeek API context caching
• DeepSeek API