AIVory Smart Inference · DeepSeek

Open Weight

DeepSeek V4 Flash - fast inference, flat-rate tokens.

DeepSeek's speed-first MoE model. Optimized for throughput over peak benchmark scores. 128K context, low per-token cost.

Start using DeepSeek V4 Flash See pricing

At a glance

DeepSeek V4 Flash via Smart Inference.

Routed to the cheapest live provider on every request. Same model, lower bill.

MoE

Parameters

MoE

128K

Context window

tokens per request

—

Cheapest input price

per 1M tokens via Smart Inference

—%

Cheaper

vs the priciest endpoint for the same model

Pricing

DeepSeek V4 Flash: price comparison.

Our cheapest routed price vs what you'd pay elsewhere.

Prices in USD per 1M input tokens · updated live from the router

Quick start

Use DeepSeek V4 Flash in three lines.

OpenAI-compatible. Swap your base_url, keep everything else.

from openai import OpenAI

client = OpenAI(
    base_url="https://smart.aivory.net/v1",
    api_key="sk-aivory-...",
)

resp = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Ship it."}],
)
print(resp.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://smart.aivory.net/v1",
  apiKey: process.env.AIVORY_API_KEY,
});

const resp = await client.chat.completions.create({
  model: "deepseek-v4-flash",
  messages: [{ role: "user", content: "Ship it." }],
});

console.log(resp.choices[0].message.content);

curl https://smart.aivory.net/v1/chat/completions \
  -H "Authorization: Bearer $AIVORY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Ship it."}]
  }'

That's the whole migration. Get early access and try it →

Use cases

What DeepSeek V4 Flash is great at.

Batch Processing Pipelines

Classify 50K documents, extract entities, tag sentiment. V4 Flash processes high-volume batch jobs at a lower cost-per-token than V3.2, with faster time-to-first-token. Use it when throughput matters more than peak accuracy.
Real-Time Chat at Scale

Low latency and cheap tokens make V4 Flash a good backend for user-facing chatbots. Serve thousands of concurrent conversations without the per-request cost of a frontier model.
Prompt Engineering and Experimentation

Run 500 prompt variants against a test set for under $1. V4 Flash is cheap enough to iterate quickly. Graduate to V3.2 or a larger model once you've locked down your prompt.

FAQ

Common questions.

What does DeepSeek V4 Flash cost on AIVory?

V4 Flash is one of the cheapest models per token. Exact rates depend on provider load. Check the live table above for the current input/output prices.

V4 Flash vs V3.2 - what's the tradeoff?

V4 Flash is faster and cheaper. V3.2 (671B, 37B active) is more accurate on complex reasoning, code, and math. Use V4 Flash for volume work, V3.2 for hard problems.

Is V4 Flash good enough for production?

Depends on the task. For classification, extraction, summarization, and simple chat, yes. For multi-step reasoning, code review, or tasks where accuracy is critical, test it against V3.2 or a 70B model first.

What's the architecture?

Mixture-of-experts, like V3.2 but tuned for inference speed. DeepSeek hasn't published the exact parameter count or active expert count yet. The 128K context window matches most other models in the catalogue.

Explore

More models.

DeepSeek V3.2 DeepSeek · 671B · 160K context View pricing → Gemma 2 9B Gemma · 9B · 8K context View pricing → Gemma 4 31B Gemma · 31B · 128K context View pricing → Llama 3.3 70B Llama · 70B · 128K context View pricing → All Models Browse the full catalogue with live pricing View all → Smart Inference AI inference routing · cheapest provider per request Learn more → GPU Marketplace Spot GPUs from $0.10/hr · H100, H200, RTX 4090 Compare prices →

Start using DeepSeek V4 Flash. Today.

Swap your base_url, keep shipping.

The router is live. Credits from $10, no subscription, no commitment.

Get early access Read the docs