DeepSeek-V2
Foundation Models · Free
Mixture-of-experts LLM with strong coding and reasoning
At a glance
- Made by
- DeepSeek
- Korean
- None (English and a few others only)
- API
- Yes
- Available on
- Web, API
- Open source
- Code: MIT / Model: DeepSeek Model Agreement
- GitHub
- 5,046 stars · MIT
What is DeepSeek-V2?
DeepSeek-V2 is a Mixture-of-Experts (MoE) language model that activates only 21 billion of its 236 billion total parameters for each token. It is stronger than DeepSeek 67B while cutting training costs by 42.5%, reducing the KV cache by 93.3%, and raising maximum generation throughput by 5.76 times. It was pretrained on a high-quality corpus of 8.1 trillion tokens, then went through supervised fine-tuning (SFT) and reinforcement learning. The GitHub repository provides model downloads, evaluation results, and API platform guidance. The official DeepSeek site currently runs a web chat, an app, and an API platform, and the latest model is DeepSeek-V4.1-Flash.
- MoE architecture: 236B total parameters, 21B activated per token
- 42.5% lower training cost, 93.3% smaller KV cache
- 5.76x maximum generation throughput
- Pretrained on 8.1 trillion tokens + SFT and reinforcement learning
- Model downloads available
- Official API platform (latest model: JSON output, tool calling, 1M context)
Pricing
| Plan | Price | Main limits |
|---|---|---|
| API deepseek-flash (DeepSeek-V4.1-Flash) input (cache miss) | $0.3 | Per 1 million tokens, peak $0.3 / off-peak $0.15 |
| API deepseek-flash (DeepSeek-V4.1-Flash) output | $1.2 | Per 1 million tokens, peak $1.2 / off-peak $0.6 |
| API deepseek-v4-pro input (cache miss) | $1.32 | Per 1 million tokens, peak $1.32 / off-peak $0.66 |
| API deepseek-v4-pro output | $3.96 | Per 1 million tokens, peak $3.96 / off-peak $1.98 |
Checked on the official pages on 2026-10-14. Prices and features can change, so check the official site again before paying. · Official pricing page
AI tools similar to DeepSeek-V2 (alternatives)
DeepSeek-V2 vs. alternatives at a glance
| AI | Pricing | Combos on the map | Connected tools | Free | Cheapest monthly | Korean |
|---|---|---|---|---|---|---|
| DeepSeek-V2 | Free | 0 | 16 | — | — | No |
| Grok-2 | Free + paid | 30 | 5 | Yes | $8/mo | — |
| Llama | Free | 75 | 5 | — | — | No |
| Gemini | Free + paid | 81 | 16 | Yes | — | Yes |
| Claude | Free + paid | 83 | 16 | Yes | $20/mo | Yes |
3 of these 5 tools have a free plan. Cheapest paid plan: Grok-2 ($8/mo).
Combo and connection counts are based on the relationships recorded on AI Galaxy Map (2026-10-26); they are not a quality ranking.
DeepSeek-V2 on AI Galaxy Map
- One of 60 AI tools in the Foundation Models field.
- Connected to 16 other tools — #6 in its field.
Frequently asked questions
Can I use DeepSeek-V2 for free?
API deepseek-flash (DeepSeek-V4.1-Flash) input (cache miss) $0.3, API deepseek-flash (DeepSeek-V4.1-Flash) output $1.2, API deepseek-v4-pro input (cache miss) $1.32, API deepseek-v4-pro output $3.96. Prices can change, so check the official site before paying.
What AI can I use instead of DeepSeek-V2?
Tools that do the same job, such as Grok-2, Llama, Gemini, Claude, are connected to it on the map.
Can I use DeepSeek-V2 in Korean?
None (English and a few others only)
Sources (official pages)
- https://deepseek.com/en/
- https://github.com/deepseek-ai/DeepSeek-V2
- https://api-docs.deepseek.com/quick_start/pricing/
Checked on the official pages on 2026-10-14. Prices and features can change, so check the official site again before paying.