DeepSeek V4 Promotional Pricing Ends May 31: What Happens to Your AI Coding Bill
May 27, 2026 · 5 min read
The Deadline Is This Week
DeepSeek's promotional pricing for V4 Pro expires on May 31, 2026. If you are currently using deepseek-reasoner (now deepseek-v4-pro) at the promotional rate, your token costs will change after that date. This is worth calculating before it happens.
The promotional pricing was introduced alongside the rebranding of DeepSeek's API models. What was previously called deepseek-chat and deepseek-reasoner has been renamed to deepseek-v4-flash and deepseek-v4-pro respectively. Both old names are now deprecated and will eventually stop accepting requests.
Current vs. Post-Promotion Pricing
| Model | Input (cache miss) — now | Input (cache hit) — now | Output — now |
|---|---|---|---|
| deepseek-v4-flash | $0.14 / 1M | $0.0028 / 1M | $0.28 / 1M |
| deepseek-v4-pro (promotional) | $0.435 / 1M | $0.003625 / 1M | $0.87 / 1M |
DeepSeek has not published the post-promotion V4 Pro rates as of this writing. Based on the historical pattern when V3 promotional pricing ended, expect a rate increase of 40–60% on input tokens for V4 Pro. If you are heavy on output tokens (reasoning-heavy tasks), that component may increase proportionally as well.
Which Developers Are Affected Most
The pricing change matters most if you are using deepseek-v4-pro (the reasoning variant) for complex tasks: architecture planning, multi-file refactors, algorithm design, or anything that benefits from extended thinking. These tasks generate more output tokens and often involve large context windows where the cache-miss input rate applies.
For developers using deepseek-v4-flash for routine coding tasks (autocomplete, small fixes, documentation), the promotional deadline does not affect you — V4 Flash pricing is not under a time-limited promotion.
Three Actions Before May 31
1. Audit your model distribution. Check your DeepSeek API usage dashboard and see what percentage of your tokens are going to V4 Pro vs V4 Flash. If V4 Pro is under 20% of your total spend, the promotion ending will have a small overall impact.
2. Consider a temporary V4 Flash substitution. For any tasks currently routed to V4 Pro that are not strictly reasoning-dependent, test routing them to V4 Flash before the deadline. The quality gap is real but narrower than the price gap for many coding tasks.
3. Update your model names in API calls. Regardless of pricing, deepseek-chat and deepseek-reasoner are deprecated. If you are still using these identifiers in your code or agent configuration, migrate to deepseek-v4-flash and deepseek-v4-pro before they stop accepting requests.
The Broader Context
Promotional pricing periods have become a standard part of the AI model launch playbook. DeepSeek, Anthropic, and OpenAI all use time-limited discounts to drive adoption of new model versions. Developers who build cost assumptions into product budgets based on promotional rates are exposed to mid-cycle price resets that can meaningfully change unit economics.
The safest approach: model your AI costs at 1.5× the current promotional rate so that expiry events are absorbed without budget surprises. Use the AI Cost Estimator to run these scenarios before committing to a model choice for a new project.
Want to calculate exact costs for your project?
Related Articles
GPT-5.5 vs Claude Opus 4.7 vs DeepSeek V4: AI Coding Cost Comparison (May 2026)
A detailed cost comparison of GPT-5.5, Claude Opus 4.7, and DeepSeek V4 for AI-assisted coding. See exactly how much each model costs for real development tasks.
How DeepSeek’s Cache Pricing Changes the Real Cost of AI Coding Agents
DeepSeek V4 pricing and cache-hit economics show why repeated context, repository analysis, and long agent sessions can become much cheaper when caching works.
AMD MI355X Beats NVIDIA B200 on DeepSeek Inference Cost: What It Means for API Prices
AMD's MI355X hardware delivers DeepSeek-R1 inference at $0.169 per million tokens — 5% cheaper than NVIDIA B200 and 40% cheaper in some SGLang configurations. Here is what hardware competition means for your API bill.