---
last_updated: 2026-11-20
source_url: https://opper.ai/deepseek/deepseek-v3
title: "DeepSeek V3 by DeepSeek"
description: "DeepSeek V3, 671B sparse MoE activating 37B per token for balanced open-weight reasoning and coding."
---

# DeepSeek V3

By [DeepSeek](https://opper.ai/deepseek).

Versions: [DeepSeek V3 (0324)](https://opper.ai/deepseek/deepseek-v3-0324) · DeepSeek V3 (this page)

Released 26 Dec 2024 · [Announcement](https://api-docs.deepseek.com/news/news1226) · [All DeepSeek releases](https://opper.ai/model-releases/deepseek)

DeepSeek-V3 is a 671-billion-parameter open-weight Mixture-of-Experts model released in December 2024 that activates just 37 billion parameters per token for efficient inference. It pairs Multi-head Latent Attention with sparse expert routing to deliver strong reasoning, mathematics, and code generation.

The model was trained on 14.8 trillion tokens for roughly 5.6 million dollars in compute, a fraction of what comparable closed models reportedly required. It supports function calling and tool use, making it a practical base for production agentic workflows.

DeepSeek-V3 targets cost-conscious teams that want capable general-purpose performance without proprietary lock-in, and its architecture became the foundation for the wider V3 family of releases.

## Available routes

DeepSeek V3 runs on 2 different routes through the Opper gateway. Prices are per 1M tokens. One route also sells flex processing: the same model at another price, picked per request with service_tier. [How service tiers work](https://docs.opper.ai/build/gateway/service-tiers). Its routes and prices, live from the catalogue, are public JSON with no API key needed at https://api.opper.ai/v3/models?q=deepinfra%2Fdeepseek-ai%2FDeepSeek-V3.

| Provider | Model id | Region | Zero data retention | Training | Input | Output | Cache read |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [DeepInfra](https://opper.ai/provider/deepinfra) | `deepinfra/deepseek-ai/DeepSeek-V3` | US (United States) | Yes | No | $0.32 | $0.89 | Input price |
| [DeepInfra](https://opper.ai/provider/deepinfra), Flex | `deepinfra:flex/deepseek-ai/DeepSeek-V3` | US (United States) | Yes | No | $0.26 | $0.71 | $0.26 |
| [Novita](https://opper.ai/provider/novita) | `novita/deepseek/deepseek_v3` | US (United States) | Yes | No | $0.89 | $0.89 | Input price |

## Key info

- Model id: `deepinfra/deepseek-ai/DeepSeek-V3`
- Input: Text
- Output: Text
- Features: Tools
- Context window: 164K
- Max output: 164K
- Input price: $0.32 per 1M tokens
- Output price: $0.89 per 1M tokens
- Released: 26 Dec 2024
- US residency available
- Zero data retention on pay-as-you-go
- No training by default

## Data handling per route

Each route hosting DeepSeek V3 has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.

### DeepInfra — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: No classifier
- Transfer mechanism: SCCs

Verified by Opper on 2026-10-19.

### Novita — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: On by default, keeps nothing
- Transfer mechanism: SCCs

Verified by Opper on 2026-09-15.

## Best for

Where DeepSeek V3 ranks among the models on Opper with a score for each use case (0 to 100).

- [Agents](https://opper.ai/models?use=agents): 0.3 (Terminal-Bench 4.0 and AutomationBench)

## Benchmarks

Independent benchmark scores via [Artificial Analysis](https://artificialanalysis.ai) ([view on AA](https://artificialanalysis.ai/models/deepseek-v3)). Indices and domain scores are out of 100, Omniscience runs from −100 to 100, Briefcase is an Elo rating, and the other evaluation scores are the share of tasks solved.

- Global rank: #320 of 520 LLMs on the [LLM leaderboard](https://opper.ai/llm-leaderboard)
- Tier: Efficient
- Output speed: 84 tok/s
- Intelligence Index (previous version): 8.5

### Coding

- Terminal-Bench 4.0: 0%
- SciCode: 36%
- LiveCodeBench: 36%

### Agentic & knowledge work

- AutomationBench: 1%
- GDPval: 0%
- GDP.pdf: 1%
- Briefcase (Elo): 78
- Terminal-Bench Hard: 7%
- τ²-Bench Telecom: 23%

### Science & reasoning

- Humanity's Last Exam: 3%
- GPQA Diamond: 56%
- Long-context reasoning: 29%

### Knowledge

- MMLU-Pro: 75%

### Math & instruction following

- AIME 2025: 26%
- IFBench: 35%

## Get started

Call DeepSeek V3 through the Opper gateway with one API key. Opper is drop-in compatible with the OpenAI, Anthropic and Google AI SDKs.

DeepSeek V3 is a premium model on Opper. Sign up needs no credit card: you get an API key straight away and the free models work in the playground and the API. Add a card to use premium models, pay-as-you-go with no minimum.

### Set it up with your agent

Paste this into a coding agent like Claude Code, Cursor or Codex and it will wire up Opper for you:

```text
Use curl to download, read and follow: https://skills.opper.ai Call DeepSeek V3
```

### Or call it directly

OpenAI:

```javascript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const completion = await client.chat.completions.create({
  model: "deepinfra/deepseek-ai/DeepSeek-V3",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(completion.choices[0].message.content);
```

Anthropic:

```javascript
import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const message = await client.messages.create({
  model: "deepinfra/deepseek-ai/DeepSeek-V3",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(message.content);
```

Gemini:

```javascript
import { GoogleGenAI } from "@google/genai";

const ai = new GoogleGenAI({
  apiKey: process.env.OPPER_API_KEY,
  httpOptions: { baseUrl: "https://api.opper.ai/v3/compat" },
});

const response = await ai.models.generateContent({
  model: "deepinfra/deepseek-ai/DeepSeek-V3",
  contents: "Hello",
});
console.log(response.text);
```

cURL:

```bash
curl https://api.opper.ai/v3/compat/chat/completions \
  -H "Authorization: Bearer $OPPER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepinfra/deepseek-ai/DeepSeek-V3",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'
```

## Use DeepSeek V3 in your apps

Pick your tool to see how to run it on DeepSeek V3 through Opper.

- [Claude Code](https://opper.ai/apps/claude-code/deepseek-v3)
- [Claude Cowork](https://opper.ai/apps/claude-cowork/deepseek-v3)
- [Codex](https://opper.ai/apps/codex/deepseek-v3)
- [DeepSeek Harness](https://opper.ai/apps/deepseek-harness/deepseek-v3)
- [GitHub Copilot CLI](https://opper.ai/apps/github-copilot-cli/deepseek-v3)
- [Aider](https://opper.ai/apps/aider/deepseek-v3)
- [Goose](https://opper.ai/apps/goose/deepseek-v3)
- [Droid](https://opper.ai/apps/droid/deepseek-v3)
- [OpenCode](https://opper.ai/apps/opencode/deepseek-v3)
- [Pi](https://opper.ai/apps/pi/deepseek-v3)
- [OpenClaw](https://opper.ai/apps/openclaw/deepseek-v3)
- [Hermes](https://opper.ai/apps/hermes/deepseek-v3)
- [Copilot for Obsidian](https://opper.ai/apps/obsidian-copilot/deepseek-v3)
- [OmniRoute](https://opper.ai/apps/omniroute/deepseek-v3)
- [Cursor](https://opper.ai/apps/cursor/deepseek-v3)
- [Cline](https://opper.ai/apps/cline/deepseek-v3)
- [Continue](https://opper.ai/apps/continue/deepseek-v3)
- [Kilo Code](https://opper.ai/apps/kilo-code/deepseek-v3)
- [GitHub Copilot](https://opper.ai/apps/github-copilot/deepseek-v3)
- [Zed](https://opper.ai/apps/zed/deepseek-v3)
- [JetBrains AI Assistant](https://opper.ai/apps/jetbrains-ai-assistant/deepseek-v3)
- [Voidleap Code](https://opper.ai/apps/voidleap-code/deepseek-v3)
- [n8n](https://opper.ai/apps/n8n/deepseek-v3)
- [LibreChat](https://opper.ai/apps/librechat/deepseek-v3)
- [Open WebUI](https://opper.ai/apps/open-webui/deepseek-v3)
- [See all apps](https://opper.ai/apps)

## Other models from DeepSeek

- [DeepSeek V4.1 Flash (1017)](https://opper.ai/deepseek/deepseek-v4-1-flash-1017)
- [DeepSeek V4 Pro](https://opper.ai/deepseek/deepseek-v4-pro)
- [DeepSeek V4 Pro 0813](https://opper.ai/deepseek/deepseek-v4-pro-0813)
- [DeepSeek V4.1 Flash](https://opper.ai/deepseek/deepseek-v4-1-flash)
- [DeepSeek V4 Flash (0731)](https://opper.ai/deepseek/deepseek-v4-flash-0731)
- [DeepSeek V4 Flash](https://opper.ai/deepseek/deepseek-v4-flash)
