DeepSeek
The open-weight family everyone else gets benchmarked against.
DeepSeek publishes the weights for nearly everything it ships, which is why its models became the price anchor for the whole market — a Flash model here costs a fraction of a frontier model and gets close on reasoning and code. The V4 line is a sparse mixture-of-experts design with a million-token context. This row is the model family; DeepSeek's own free chat app has its own entry in Chats.
At a glance — DeepSeek V3.2
- Cost
- $0.269 in · $0.40 out per 1M tokensBudget
- Memory
- 164K tokens — about 250 pages
- Reads
- Text
- Can do
- Use tools (agents)
- Structured output
- Thinking mode
- Good at
- Coding: better than 45%
- Reliability
- 100.00% uptime, last 24h
- Released
- Dec 1, 2025
Advanced details
- API model ID
deepseek/deepseek-v3.2- Max output
- 66K tokens
- Max input
- 98K tokens
- Cached input
- $0.134 per 1M
- Tokenizer
- DeepSeek
- Coding index
- 44.2
Providers
The same model, hosted by different companies. OpenRouter picks one per request and falls back to the next if it fails.
| Provider | Input /1M | Output /1M | Cached /1M | Max output | Uptime 24h |
|---|---|---|---|---|---|
| GMICloud · fp8 | $0.209 | $0.31 | $0.022 | 147K | 98.22% |
| SiliconFlow · fp8 | $0.259 | $0.42 | $0.135 | 147K | 95.14% |
| DeepInfra · fp4 | $0.26 | $0.38 | $0.13 | 16K | 97.51% |
| AtlasCloud · fp8 | $0.26 | $0.38 | $0.13 | 147K | 96.31% |
| Venice | $0.268 | $0.39 | $0.13 | 33K | 98.09% |
| Novita · fp8 | $0.269 | $0.40 | $0.134 | 66K | 99.98% |
| Baidu · fp8 | $0.28 | $0.42 | $0.028 | 66K | 100.00% |
| DigitalOcean | $0.30 | $0.96 | $0.09 | 147K | 98.57% |
| Alibaba · fp8 | $0.371 | $1.112 | $0.074 | 66K | 95.83% |
| Friendli | $0.50 | $1.50 | $0.25 | 147K | 99.99% |
| $0.56 | $1.68 | — | 66K | 99.31% | |
| Phala | $1.00 | $1.00 | $0.50 | 64K | 99.29% |
| Mara | $3.00 | $4.50 | — | 7K | 78.00% |
| SambaNova | $3.00 | $4.50 | — | 7K | 98.33% |
Supported parameters
- frequency_penalty
- include_reasoning
- logit_bias
- logprobs
- max_tokens
- min_p
- presence_penalty
- reasoning
- repetition_penalty
- response_format
- seed
- stop
- structured_outputs
- temperature
- tool_choice
- tools
- top_k
- top_logprobs
- top_p
Live from OpenRouter, refreshed hourly. Scores from Artificial Analysis.
Key info
- Pricing
- Open source
- Category
- Models
- Best for
- Intermediate
- Made by
- DeepSeek
- Weights
- Open
- Good for
- Reasoning and code at a fraction of frontier prices