We send our hardest tickets to the DeepSeek V4 Pro API and keep the volume on Flash.
DeepSeek V4 Pro is the flagship, high-capability tier of the newest DeepSeek generation — the model you reach for on complex reasoning, hard coding, and multi-step agents, when getting it right beats shaving a few cents. Served on RouterBase over an OpenAI-compatible endpoint at $0.435 / 1M input and $0.87 / 1M output, with cached input at just $0.003625; route everyday volume to DeepSeek V4 Flash and escalate here for the hardest jobs.
Input
DeepSeek V4 Pro
Hi! I'm a helpful AI assistant. What can I do for you?
DeepSeek V4 Pro API — The Top of the DeepSeek V4 Line
The DeepSeek V4 Pro API serves DeepSeek V4 Pro, the high-capability tier of the newest DeepSeek generation — the one you reach for on the hardest work.
DeepSeek V4 Pro is the flagship tier of the newest DeepSeek generation: the model you reach for when a job needs the most capability, not the lowest latency. The DeepSeek V4 Pro API is built for complex reasoning, hard coding, and multi-step agents, where getting the answer right matters more than shaving a few cents. It bills $0.435 per 1M input tokens and $0.87 per 1M output, with cached input at just $0.003625 — priced above DeepSeek V4 Flash for the extra capability.
You reach the DeepSeek V4 Pro API through RouterBase on the OpenAI chat-completions protocol; set the model to deepseek-v4-pro and your client is unchanged. Route everyday volume to DeepSeek V4 Flash and escalate to the DeepSeek V4 Pro API when the task is hard — one key spans both, and the whole catalog.
What the DeepSeek V4 Pro API gives you
Reach for it when the work is hard.
Top of the V4 line
The DeepSeek V4 Pro API is the high-capability tier of the newest DeepSeek generation.
Built for hard work
Complex reasoning, hard coding, and multi-step agents are where the DeepSeek V4 Pro API earns its place.
Quality over pennies
When getting the answer right matters more than shaving cost, the DeepSeek V4 Pro API is the call.
Near-free cached reads
Cached input is just $0.003625 per 1M, so repeated context on the DeepSeek V4 Pro API costs almost nothing.
Coding and agents
Strong code plus reliable function calling make the DeepSeek V4 Pro API a solid backbone for demanding assistants and agents.
Pair with V4 Flash
Route volume to DeepSeek V4 Flash and escalate to the DeepSeek V4 Pro API for the hardest jobs — one key covers both.
Get going
First call to the DeepSeek V4 Pro API in three steps
Wire it once, then route the hard work to it.
Create a key
One RouterBase key reaches the DeepSeek V4 Pro API and every other model in the catalog.
Point and name
Send any OpenAI client to routerbase.com/v1 with model=deepseek/deepseek-v4-pro. Streaming and function calling work out of the box.
Route by difficulty
Send hard tasks to the DeepSeek V4 Pro API and lighter ones to V4 Flash; usage comes back per response so you can tune the split.
Pay only for what you use
RouterBase passes through partner-tier pricing. Compared against the model's official published API rate.
Teams building on the DeepSeek V4 Pro API
The flagship tier, saved for the jobs that need it.
On our toughest coding tasks, the DeepSeek V4 Pro API got answers Flash could not.
The DeepSeek V4 Pro API is where we route anything that has to be right the first time.
Cached reads at $0.003625 kept the DeepSeek V4 Pro API affordable even for long agent runs.
One key gave us Flash for volume and the DeepSeek V4 Pro API for the hard path.
Our multi-step agents got noticeably steadier on the DeepSeek V4 Pro API.
We reach for the DeepSeek V4 Pro API when correctness beats saving a few cents.
Switching hard prompts to the DeepSeek V4 Pro API was one string, and the quality jump was obvious.
Function calling held up, so our agent stack runs the DeepSeek V4 Pro API on its toughest steps.
DeepSeek V4 Pro API — common questions
What to know before you route traffic to it.
RouterBase's hosted access to DeepSeek V4 Pro, the high-capability tier of the newest DeepSeek generation, over an OpenAI-compatible endpoint.