ReasoningMoEContext: 125K
DeepSeek V3
Comprehensive deployment profile and benchmark telemetry for DeepSeek V3. Self-hosting memory footprints, verified token economics, and direct API endpoints.
Verified Engineering Benchmarks
SWE-bench
45.2%
LiveCodeBench
61.6%
MATH-500
95.8%
MMLU
88.5%
VRAM Requirements & Sizing
🔗 Quick Actions
🖥️ Compatible GPUs & Self-Host Pricing
Deterministic VRAM math from entity graph. Green = fits in single GPU.
| GPU | VRAM | FP16 | FP8 | INT4 | Calculator Link |
|---|---|---|---|---|---|
| H100 SXM5 | 80 GB | OOM | OOM | OOM | Pre-filled → |
| H200 | 141 GB | OOM | OOM | OOM | Pre-filled → |
| B200 | 192 GB | OOM | OOM | OOM | Pre-filled → |
| A100 80GB | 80 GB | OOM | OOM | OOM | Pre-filled → |
| L40S | 48 GB | OOM | OOM | OOM | Pre-filled → |
| RTX 4090 | 24 GB | OOM | OOM | OOM | Pre-filled → |
Active API Providers & Free Tiers
| Provider | Input / 1M | Output / 1M | Speed | Free Tier Status |
|---|---|---|---|---|
| DeepSeek | $0.14 | $0.28 | 35 tok/s | Paid API Only |
| OpenRouter | $0.14 | $0.28 | 30 tok/s | Paid API Only |
OpenAI SDK Drop-in
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.deepseek.com/v1",
apiKey: process.env.DEEPSEEK_API_KEY,
});
const response = await client.chat.completions.create({
model: "deepseek-v3",
messages: [{ role: "user", content: "Hello" }],
});