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 V4 Flash 0423
- Cost
- $0.047 in · $0.094 out per 1M tokensBudget
- Memory
- 1M tokens — about 1,550 pages
- Reads
- Text
- Can do
- Use tools (agents)
- Structured output
- Thinking mode
- Good at
- Coding: better than 65% · Agentic work: better than 56% · Overall: better than 46%
- Reliability
- 99.97% uptime, last 24h
- Released
- Apr 24, 2026
Advanced details
- API model ID
deepseek/deepseek-v4-flash- Max output
- 384K tokens
- Max input
- 1M tokens
- Cached input
- $0.009 per 1M
- Tokenizer
- DeepSeek
- Coding index
- 56.2
- Agentic work index
- 22.2
- Overall index
- 24.2
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 |
|---|---|---|---|---|---|
| Relace · fp4 | $0.03 | $0.50 | $0.016 | 944K | 99.54% |
| OpenInference · fp8Degraded | $0.04 | $0.50 | $0.014 | 944K | 96.34% |
| Baidu · fp8 | $0.047 | $0.094 | $0.009 | 131K | 97.58% |
| StreamLake · fp8 | $0.047 | $0.094 | $0.009 | 384K | 95.81% |
| DeepInfra · fp8 | $0.09 | $0.18 | $0.018 | 66K | 99.75% |
| GMICloud · fp8 | $0.091 | $0.182 | $0.018 | 944K | 99.47% |
| Venice | $0.097 | $0.193 | $0.02 | 33K | 99.12% |
| DigitalOcean | $0.098 | $0.196 | $0.02 | 384K | 99.85% |
| SiliconFlow · fp8 | $0.13 | $0.28 | $0.028 | 393K | 99.50% |
| Alibaba · fp8 | $0.134 | $0.268 | $0.027 | 393K | 98.05% |
| Novita · fp8 | $0.14 | $0.28 | $0.028 | 393K | 99.97% |
| AtlasCloud · fp4 | $0.14 | $0.28 | $0.028 | 393K | 98.79% |
| Parasail · fp8 | $0.14 | $0.28 | $0.07 | 944K | 99.77% |
| NextBit · fp8 | $0.15 | $0.30 | $0.035 | 944K | 99.84% |
| Mancer 2 · fp8 | $0.19 | $0.50 | — | 944K | 97.75% |
| Azure · us | $0.21 | $0.56 | $0.031 | 384K | 97.96% |
Supported parameters
- frequency_penalty
- include_reasoning
- logit_bias
- logprobs
- max_completion_tokens
- max_tokens
- min_p
- presence_penalty
- reasoning
- reasoning_effort
- repetition_penalty
- response_format
- seed
- stop
- structured_outputs
- temperature
- tool_choice
- tools
- top_a
- 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