Key details

  1. DeepSeek V4 Pro is generally available as DeepSeek-V4-Pro-0813.
  2. The model page lists a 1M-token context window and up to 384K output tokens.
  3. Peak windows are 01:00–04:00 UTC and 06:00–10:00 UTC.
  4. Peak rates per million tokens are $0.044 cached input, $1.32 uncached input and $3.96 output.
  5. Off-peak rates are exactly 50% lower: $0.022, $0.66 and $1.98 respectively.
  6. DeepSeek lists a concurrency limit of 500 for V4 Pro.
  7. Reuters reported that the new time-based pricing took effect on August 17, 2026.

What builders should take away

  1. Classify inference jobs as latency-sensitive or deferrable, then move queues, evaluations and batch generation outside DeepSeek's two UTC peak windows where deadlines allow.
  2. Convert the UTC peak windows into explicit scheduler rules and monitor them centrally rather than relying on developers to remember the price schedule.
  3. Track cached input, uncached input and output separately; a workload with high prefix reuse can have very different economics from one dominated by uncached context.
  4. Benchmark your own agent or coding tasks at measured success rate, latency and total cost before routing production traffic on the strength of aggregate model rankings alone.

What changed

DeepSeek has made V4 Pro generally available and is charging two sets of API rates depending on time of day. The current pricing page lists peak windows of 01:00–04:00 UTC and 06:00–10:00 UTC; outside those windows, cached input, uncached input and output are each priced at half the peak rate. The production model is listed as DeepSeek-V4-Pro-0813 with a 1 million-token context window and 384,000-token maximum output.

Why it matters

Most model-price comparisons treat token rates as static. DeepSeek's schedule makes latency tolerance economically valuable: batch jobs, indexing, evaluations, report generation and other deferrable work can materially reduce token spend simply by moving outside peak windows. It also makes cache-hit rate and output volume more important to routing decisions because each component has a separate meter.

The same model has two daily price regimes

DeepSeek lists V4 Pro at peak rates of $0.044 per million cached input tokens, $1.32 per million uncached input tokens and $3.96 per million output tokens. Off-peak rates are $0.022, $0.66 and $1.98 respectively. The two peak windows total seven hours per day, leaving the remaining 17 hours at the lower rates. The pricing page says billing uses the prevailing rate when a request is processed.

Scheduling becomes a routing primitive

For latency-sensitive requests, time-of-day pricing may be irrelevant. For queues and background work, however, a scheduler can classify jobs by deadline and defer eligible work until off-peak periods. Builders comparing DeepSeek with other providers should therefore model effective cost by workload timing rather than comparing only a single headline token price.

Caching still changes the economics substantially

The cache-hit rate is far below uncached-input pricing in both periods, so systems that repeatedly send long shared prefixes, tool definitions or context can benefit from both caching and off-peak execution. Cost monitoring should keep cached input, uncached input and generated output separate rather than collapsing everything into one blended token rate.

Performance context still needs workload-level testing

Independent reporting places V4 Pro among stronger current reasoning and coding models, but benchmark summaries do not establish that it is the best choice for a particular agent, language, latency target or tool-use workload. The useful comparison is task success at a measured cost and latency under the exact API configuration a builder will run.

Timeline

2026-08-13

DeepSeek announces V4 Pro

DeepSeek released V4 Pro as its production V4 model.
2026-08-17

Peak/off-peak pricing takes effect

DeepSeek's time-dependent API pricing became effective, with off-peak rates set at half peak rates.

What to watch next

  • Whether DeepSeek changes the peak windows or rate ratio as demand shifts.
  • Independent agent, coding and long-context evaluations using the production V4 Pro endpoint.
  • Whether other API providers adopt time-of-day pricing or queue-aware discounts for deferrable inference.

Still unclear

  • Independent production evidence for V4 Pro is still limited relative to the new model's age; vendor and aggregate benchmark results may not predict a specific workload.
  • API prices, concurrency limits and peak windows are operational terms that DeepSeek can change, so production schedulers should not hard-code them without monitoring documentation.

Sources

Direct reading behind this dossier.

3 sources
Models & Pricing
DeepSeek primary

Current model lineup and token pricing for V4 Flash, V4 Pro and V4 Flash Vision-Exp.