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 0731
- Cost
- $0.022 in · $0.32 out per 1M tokensBudget
- Memory
- 1.3M tokens — about 1,950 pages
- Reads
- Text
- Can do
- Use tools (agents)
- Structured output
- Thinking mode
- Good at
- Coding: better than 75% · Agentic work: better than 76% · Overall: better than 66%
- Reliability
- 99.99% uptime, last 24h
- Released
- Jul 31, 2026
Advanced details
- API model ID
deepseek/deepseek-v4-flash-0731- Max output
- 944K tokens
- Cached input
- $0.016 per 1M
- Tokenizer
- DeepSeek
- Coding index
- 69.1
- Agentic work index
- 41.0
- Overall index
- 34.3
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 | $0.022 | $0.32 | $0.016 | 944K | 99.82% |
| Sail Research · us | $0.023 | $0.42 | $0.012 | 944K | 98.86% |
| Sail Research · fp4 | $0.023 | $0.42 | $0.012 | 944K | 98.59% |
| OpenInference · fp8 | $0.03 | $0.40 | $0.008 | 944K | 98.81% |
| Inceptron · fp4 | $0.04 | $0.271 | $0.03 | 944K | 99.94% |
| StreamLake · fp8 | $0.044 | $0.132 | $0.001 | 384K | 99.71% |
| Baidu · fp8 | $0.06 | $0.18 | $0.002 | 131K | 99.95% |
| DeepInfra · fp8 | $0.06 | $0.18 | $0.015 | 384K | 99.89% |
| Wafer · fast | $0.07 | $0.35 | $0.06 | 944K | 99.98% |
| Reka | $0.088 | $0.528 | $0.006 | 131K | 99.72% |
| Makora | $0.09 | $0.195 | $0.02 | 384K | 99.43% |
| DigitalOcean | $0.119 | $0.238 | $0.024 | 944K | 99.85% |
| BaseTen · fp8 | $0.13 | $0.26 | $0.028 | 384K | 99.64% |
| BaseTen · fp8 | $0.13 | $0.26 | $0.028 | 384K | 99.65% |
| CoreWeave · fp8 | $0.13 | $0.28 | $0.07 | 236K | 99.99% |
| Nebius · fp8 | $0.14 | $0.28 | — | 922K | 94.28% |
| Cohere | $0.14 | $0.28 | $0.07 | 384K | 99.01% |
| Together | $0.14 | $0.28 | $0.03 | 944K | 99.61% |
| Parasail · fp8 | $0.14 | $0.28 | $0.05 | 944K | 99.79% |
| Morph | $0.142 | $0.40 | $0.036 | 944K | 99.89% |
| Venice | $0.175 | $0.35 | $0.035 | 33K | 99.82% |
| Mancer 2 · fp8 | $0.20 | $0.60 | — | 944K | 99.51% |
| Fireworks | $0.22 | $0.66 | $0.007 | 944K | 98.27% |
| SiliconFlow · fp8 | $0.22 | $0.66 | $0.028 | 393K | 99.68% |
| GMICloud · fp8 | $0.286 | $0.858 | $0.009 | 944K | 99.51% |
| Phala | $0.308 | $0.924 | $0.02 | 393K | 99.39% |
| NextBit · fp8 | $0.352 | $1.056 | $0.012 | 944K | 99.99% |
| Novita · fp8 | $0.409 | $1.228 | $0.026 | 393K | 99.99% |
| AtlasCloud · fp4 | $0.44 | $1.32 | $0.028 | 393K | 97.87% |
| Cloudflare | $0.44 | $1.32 | $0.014 | 1.2M | 99.99% |
Supported parameters
- frequency_penalty
- include_reasoning
- logit_bias
- logprobs
- max_tokens
- min_p
- parallel_tool_calls
- 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