deepseek-v4-pro
Input price$0.66
Output price$1.98
DeepSeekText
deepseek-reasoner-v4
deepseek-reasoner-v4 is DeepSeek's model at $0.43 input and $0.87 output per million tokens, with a 131K context window. On a typical 3:1 input-to-output mix that blends to roughly $0.54 per million tokens, placing it in the mid-range band of the 89 models tracked here — cheaper than 64% of them.
Context window131K
Input price$0.435 / 1M
Output price$0.87 / 1M
Output speed30 tok/s
Quality score92
Data updatedUpdated Aug 2026
Price verified2026-08-16 · Official
Price noteOff-peak list rate; peak hours (01:00–04:00 and 06:00–10:00 UTC) bill at 2×
By price, deepseek-reasoner-v4 ranks 36th cheapest of 89 on input and 26th on output. Its quality score of 92 ranks 19th of 89, and on value (quality score ÷ output price) it ranks 22nd of 89. Its blended rate sits 82% below the catalog average.
In real billing terms: a support chatbot handling 50,000 turns a month (2,000 input / 500 output tokens each) costs about $65 per month on deepseek-reasoner-v4, while a RAG workload of 20,000 queries (8,000 input / 800 output tokens each) runs about $84. Because output is priced 2.0× input here, long-generation tasks scale cost fastest — cap output length before you optimise anything else.
A 131K context window holds roughly 175 pages of A4 text (at ~750 tokens per page), ranking 68th of 89 here. That covers typical RAG and multi-turn chat, though book-length inputs still need retrieval and chunking. Output runs at about 30 tok/s (79th fastest of 89), which is better suited to asynchronous or background jobs.
We track 4 DeepSeek models, and deepseek-reasoner-v4's blended rate sits 3% below that provider's average. If your task tolerates a little less headroom, qwen3.5 (Alibaba) blends about 79% cheaper at a quality score of 91, and is worth benchmarking first. Going the other way, deepseek-v4-pro scores 93 on quality for about 82% more.
The closest model in DeepSeek's line-up is deepseek-chat-v4 ($0.26 input / $1.03 output, quality 88). Against it, deepseek-reasoner-v4 scores 4 points higher on quality, runs 110 tok/s slower (about 79%), costs 17% more blended. When two models sit this close on price, the deciding factor is usually the quality ceiling — run both against your real prompts before picking.
Our read on it
Quality 92 at a blended $0.54 per million tokens. The bill for that is 30 tokens/second — among the slowest we track.
Put it next to something with the same score
deepseek-reasoner-v4 is $0.435/$0.87 at quality 92. claude-sonnet-5 also scores 92 and lists at $2/$10 — 4.6x the input rate and 11.5x the output rate. That is not a few percentage points of difference. It is an order of magnitude.
Its output rate of $0.87 is lower than DeepSeek's own chat-v4 at $1.0287, even though a reasoning model emits considerably more tokens per answer. The pricing decision is deliberate: reasoning is being sold as a commodity rather than as a premium add-on.
What 30 tokens/second actually feels like
This is among the slowest rates in the catalog, rated Slow. Concretely: a 2,000-token response takes about a minute to finish streaming, where claude-haiku-4-5 at 180 tokens/second completes the same length in a little over ten seconds.
The billing detail matters more than the wait. On a reasoning model the thinking pass is itself output tokens. A reply that looks like three short paragraphs may have generated several thousand billed tokens getting there — and you waited for all of them. Nothing with a user watching the screen should call this model directly.
The right shape is a background queue
It fits deep work that can run asynchronously: code review, involved data analysis, multi-step mathematical or financial derivation, the planning stage of an agent. All of those can sit in a queue and deliver a notification rather than a stream.
Layer the peak/off-peak scheme on top — 2x during 01:00 to 04:00 and 06:00 to 10:00 UTC — and the conclusion writes itself: schedule reasoner-v4 into an off-peak batch queue and you are buying 92-point reasoning at close to the cheapest rate on this site. It is the most persuasive option on the slow-but-cheap curve right now, provided your architecture has somewhere for the waiting to happen.
Estimated monthly cost
Monthly API spend across four common workloads, next to the cheapest model we track.
| Workload | Usage assumption | This model | Cheapest model |
|---|---|---|---|
| Support chatbot | 2,000 input / 500 output tokens per turn, 50,000 turns per month | $65 | $3.00 |
| RAG document Q&A | 8,000 input / 800 output tokens per query, 20,000 queries per month | $84 | $3.84 |
| Coding assistant | 12,000 input / 2,000 output tokens per request, 10,000 requests per month | $70 | $3.20 |
| Batch summarization | 4,000 input / 400 output tokens per item, 200,000 items per month | $418 | $19 |
Derived directly from the rates on this page. Excludes prompt caching, batch discounts and free tiers.
How it compares
The alternatives worth benchmarking alongside it, from the same dataset.
| Model | Provider | Blended /1M | Quality | Price delta |
|---|---|---|---|---|
| deepseek-reasoner-v4 | DeepSeek | $0.54 | 92 | — |
| qwen3.5Cheaper option | Alibaba | $0.11 | 91 | -79% |
| deepseek-v4-proHigher quality | DeepSeek | $0.99 | 93 | +82% |
| deepseek-chat-v4Same provider | DeepSeek | $0.45 | 88 | -17% |
| qwen3.7-flashBest value overall | Alibaba | $0.055 | 82 | -90% |
Best for
ReasoningCoding
Modalities
Text
Where it fits — and where it doesn't
Good fit
- Coding: writing, refactoring and debugging across multi-file project context.
- Reasoning: multi-step decomposition, planning and tasks that need rigorous inference.
Poor fit
- Latency-critical interactive use: 30 tok/s makes long responses feel sluggish.
- Image understanding: this model is text-only, so vision tasks need a multimodal alternative.
Frequently asked questions
How much does the deepseek-reasoner-v4 API cost?
deepseek-reasoner-v4 costs $0.43 per million input tokens and $0.87 per million output tokens. On a 3:1 input-to-output mix that blends to about $0.54 per million tokens.
What does deepseek-reasoner-v4 cost per month?
It depends on volume. Using the four worked scenarios on this page: a support chatbot runs about $65/month, RAG document Q&A about $84/month, a coding assistant about $70/month, and batch summarization about $418/month. Plug your own token counts into the cost calculator for a tailored figure.
How large is the deepseek-reasoner-v4 context window?
131K tokens — roughly 175 pages of A4 text at ~750 tokens per page, ranking 68th of the 89 models tracked here.
deepseek-reasoner-v4 or qwen3.5 — which is better value?
qwen3.5 blends to about $0.11 per million tokens versus $0.54 for deepseek-reasoner-v4 — roughly 79% cheaper — at a quality score of 91 against 92. If that quality gap does not show up on your task, take the cheaper one; if it does, stay with deepseek-reasoner-v4. Benchmark both on your own data before deciding.
What is deepseek-reasoner-v4 best suited for?
Tagged use cases: reasoning, coding. Coding: writing, refactoring and debugging across multi-file project context. Less suitable: Latency-critical interactive use: 30 tok/s makes long responses feel sluggish.
How current is this deepseek-reasoner-v4 pricing?
Pricing here was last verified on 2026-08-16 against official documentation. llmprice.app runs a daily collection job, but providers can change rates between runs — confirm against DeepSeek's official pricing page before you commit.
Estimate the real cost of this model
Bring tokens per request and monthly volume into the calculator to compare total model costs.
Related models
Continue by provider and use case.
View model
deepseek-v4-flash
Input price$0.15
Output price$0.6
deepseek-chat-v4
Input price$0.2574
Output price$1.0287
Pricing, speed and quality scores change over time. Confirm official documentation before production use.