DeepSeek V4-Pro 75% Off Promo (Ends May 31, 2026)

DeepSeek V4-Pro is 75% off through 2026-05-31, which drops its official price to $0.435 per million input tokens and $0.87 per million output, against a list price of $1.74 / $3.48 that returns on June 1. The promo is the cheapest window you will get on the Pro tier this year. But we called it live, and the catch is not the price: V4-Pro answered in 2.1 seconds where V4-Flash takes about 0.7. The discount is real; so is the latency you trade for the quality.

Most promo coverage stops at the headline percentage. The decision that actually matters is whether V4-Pro earns its place over V4-Flash on your workload, and the promo window is exactly the right time to test that while the price gap is at its narrowest.

DeepSeek V4-Pro promo pricing (verified 2026-05)

According to DeepSeek API Docs, V4-Pro is the higher-tier general-purpose model, and the rate card publishes the promo and post-promo prices side by side so you can budget for the reversion.

Sources: DeepSeek API Docs – Pricing and DeepSeek API Docs – Pricing Details USD, verified 2026-05-13.

The reversion is the part to plan for. At the promo rate, V4-Pro input runs about 3× V4-Flash input ($0.435 vs $0.14). The moment the list price returns on June 1, that gap widens to roughly 12×. Any routing logic you tune this month against the promo price will quietly get four times more expensive overnight unless you re-check it.

We called V4-Pro to see what the premium buys (first-hand evidence)

The promo price is easy to quote. What it actually delivers takes a real call, so we sent one. The full run is in our evidence pack; here is the V4-Pro result.

Two things stood out. First, deepseek-v4-pro is a distinct, real model name. The API rejects unknown names with the message that the only supported names are deepseek-v4-pro and deepseek-v4-flash, which means deepseek-chat and deepseek-reasoner are aliases that both resolve to V4-Flash. If you want the Pro tier, you must ask for it by name. Second, V4-Pro answered in 2.13 seconds against the roughly 0.7 seconds we measured for V4-Flash in the same evidence run. The Pro tier costs more per token and responds more slowly, so the case for it has to come from output quality, not from price or speed.

SiliconFlow does not honor the V4-Pro promo

According to SiliconFlow Pricing, the third-party host charges the undiscounted V4-Pro list price of $1.74 / $3.48 regardless of the official promo. Routing V4-Pro traffic through SiliconFlow this month means paying roughly 4× what you would pay calling the official endpoint directly. This is the single widest price gap between official and third-party hosting anywhere in the DeepSeek catalog right now, and it exists only because the promo applies to the official endpoint alone.

According to DeployBase, this pattern is not unique to SiliconFlow. Third-party providers track the model's list price, not its promotional price, so any temporary official discount opens a gap that the resellers do not pass through. The practical rule during a promo: call the official endpoint directly unless a compliance requirement forces you onto a US-routed provider, in which case you accept the premium knowingly.

V4-Pro promo is also a cache opportunity

Promotional pricing is a time-limited rate set below a model's published list price to drive trial before the list rate applies. For V4-Pro it runs through 2026-05-31. The promo stacks with the prompt cache: cached input on V4-Pro bills at $0.003625 per million during the promo, so a workload with a stable system prompt pays the discounted rate on the cache-miss tail and a fraction of that on the cached prefix. If you are going to test V4-Pro, structure the prompt with the stable content first to capture both discounts at once.

Choose this if you are deciding on V4-Pro

Use the promo window as your evaluation budget. Run V4-Pro against V4-Flash on your own quality benchmark now, while the price gap is only 3× and a wrong call costs little. Call the official deepseek-v4-pro endpoint directly to get the 75% discount, and keep the prompt prefix stable so the cache discount stacks on top. Budget for the 2-second-plus latency on every V4-Pro call, since it is materially slower than V4-Flash. Then make the keep-or-drop decision before June 1, because once the list price returns the model is roughly 12× the V4-Flash rate and the math changes underneath you.

FAQ

When does the DeepSeek V4-Pro promo end? It expires on 2026-05-31. From 2026-06-01, V4-Pro reverts to its list rate of $1.74 per million input tokens and $3.48 per million output.

Is V4-Pro cheaper on SiliconFlow during the promo? No. SiliconFlow charges the undiscounted list price of $1.74 / $3.48 regardless of the official promo, roughly 4× the official promo rate. Call the official endpoint directly to get the discount.

Is V4-Pro slower than V4-Flash? Yes, noticeably. In our live test V4-Pro answered in 2.13 seconds against roughly 0.7 seconds for V4-Flash. The Pro tier trades speed for output quality, so reserve it for tasks where that quality lift matters.

How do I actually call V4-Pro? Request the model name deepseek-v4-pro explicitly. The aliases deepseek-chat and deepseek-reasoner both resolve to V4-Flash, so they will not give you the Pro tier.

---

This is part of the DeepSeek API pricing hub. For how the cache discount stacks on this promo, see the cache hit discount article.

Author: Kevin Fan, Customer Success Manager at China LLM Directory, specializing in Chinese LLM ecosystem pricing. Last verified: 2026-05-13.

<!-- METADATA { "title": "DeepSeek V4-Pro 75% Off Promo: Rates Until May 31", "slug": "deepseek-v4-pro-promo", "meta_description": "DeepSeek V4-Pro is 75% off to 2026-05-31 at $0.435/$0.87 per 1M tokens; list $1.74/$3.48 returns June 1. We called it live: 2.1s latency. Promo guide May 2026.", "focus_keyword": "deepseek v4-pro promo pricing", "secondary_keywords": ["deepseek v4-pro 75 off", "deepseek v4 pro discount", "deepseek v4-pro list price", "deepseek v4-pro vs v4-flash"], "tags": ["DeepSeek", "API Pricing", "V4-Pro", "Promo"], "category": "Pricing", "cluster_id": "deepseek-api-pricing", "cluster_role": "micro", "hub_slug": "deepseek-api-pricing", "verified_until": "2026-05-31", "evidence_file": "clients/china-llm-aggregator/articles/deepseek-api-pricing-evidence.md", "author_name": "Kevin Fan", "author_title": "Customer Success Manager", "author_linkedin": "", "author_expertise": ["Chinese LLM ecosystem", "AI infrastructure pricing"], "faq_pairs": [ {"q": "When does the DeepSeek V4-Pro promo end?", "a": "It expires on 2026-05-31. From 2026-06-01, V4-Pro reverts to its list rate of $1.74 per million input tokens and $3.48 per million output."}, {"q": "Is V4-Pro cheaper on SiliconFlow during the promo?", "a": "No. SiliconFlow charges the undiscounted list price of $1.74 / $3.48 regardless of the official promo, roughly 4x the official promo rate. Call the official endpoint directly to get the discount."}, {"q": "Is V4-Pro slower than V4-Flash?", "a": "Yes, noticeably. In our live test V4-Pro answered in 2.13 seconds against roughly 0.7 seconds for V4-Flash. The Pro tier trades speed for output quality, so reserve it for tasks where that quality lift matters."}, {"q": "How do I actually call V4-Pro?", "a": "Request the model name deepseek-v4-pro explicitly. The aliases deepseek-chat and deepseek-reasoner both resolve to V4-Flash, so they will not give you the Pro tier."} ], "external_links_used": [ {"url": "https://api-docs.deepseek.com/quick_start/pricing/", "source_name": "DeepSeek API Docs – Pricing", "claim": "V4-Pro promo $0.435/$0.87 and list $1.74/$3.48"}, {"url": "https://api-docs.deepseek.com/quick_start/pricing-details-usd", "source_name": "DeepSeek API Docs – Pricing Details USD", "claim": "V4-Pro cache-hit promo rate $0.003625/M"}, {"url": "https://www.siliconflow.com/pricing", "source_name": "SiliconFlow Pricing", "claim": "SiliconFlow lists V4-Pro at undiscounted $1.74/$3.48 regardless of official promo"}, {"url": "https://deploybase.ai/articles/deepseek-v3-pricing", "source_name": "DeployBase", "claim": "Third-party providers track list price not promo price"} ], "internal_links_used": [ {"url": "/blog/deepseek-api-pricing/", "anchor_text": "DeepSeek API pricing hub", "type": "hub"}, {"url": "/blog/deepseek-cache-discount/", "anchor_text": "cache hit discount article", "type": "sibling-micro"} ], "first_hand_evidence": { "source": "deepseek-api-pricing-evidence.md Finding 5 + Finding 4", "measured": "deepseek-v4-pro live: 28 in / 60 out = $0.0000644, latency 2.13s vs V4-Flash ~0.7s; only supported model names are deepseek-v4-pro and deepseek-v4-flash", "captured": "2026-05-13" }, "images_status": "spec-only (not generated; FAL_API_KEY unset)", "images": [ {"position": "featured", "type": "generated", "prompt": "Clean editorial chart showing DeepSeek V4-Pro promo price vs list price across input and output, with a latency callout: V4-Pro 2.1s vs V4-Flash 0.7s. Bold 2026-05-31 expiry marker. Minimal background. 16:9.", "alt": "Chart comparing DeepSeek V4-Pro 75% off promo prices against post-promo list rates with a 2026-05-31 expiry and a latency callout of 2.1s for V4-Pro versus 0.7s for V4-Flash"} ] } -->