DeepSeek V3
DeprecatedDeepSeek V3 is DeepSeek's language model with a 164K context window and up to 82K output tokens, available from 11 providers, starting at $0.2 / 1M input and $0.2 / 1M output. DeepSeek's third-generation MoE LLM with 671B total parameters (37B activated per token), excelling at coding, reasoning, and tool use.
| Benchmarks | |
|---|---|
| Intelligence Index | #367 |
| Coding Index | #143 |
| Math Index | #207 |
| MMLU-Pro | #185 |
| GPQA | #375 |
| HLE | #541 |
| LiveCodeBench | #200 |
| AIME | #94 |
| IFBench | #314 |
| Time to First Token | #443 |
| SciCode | #261 |
| MATH-500 | #82 |
| AIME 2025 | #207 |
| LCR | #322 |
| TerminalBench Hard | #248 |
| TAU2 | #308 |
| Output TPS | #211 |
Capabilities
Input2/5
Text✓
Image✓
Audio·
Video·
PDF·
Output1/5
Text✓
Image·
Audio·
Video·
Embedding·
Capabilities7/13
Reasoning✓
Adaptive Reasoning·
Function Calling✓
Parallel Function Calling✓
Structured Outputs✓
Native JSON Schema✓
Web Search·
URL Context·
Computer Use·
Code Execution·
File Search·
Prompt Caching✓
Assistant Prefill✓
Pricing by Provider
US Dollar ($)
Per 1M tokens
| Provider | Standard | Flex | Fast | Batch | |||||
|---|---|---|---|---|---|---|---|---|---|
| Input $ / 1M | Output $ / 1M | Cache Read $ / 1M | Input $ / 1M | Output $ / 1M | Input $ / 1M | Output $ / 1M | Input $ / 1M | Output $ / 1M | |
| $0.58 | $1.68 | N/A | $0.29 | $0.84 | $1.01 | $2.94 | $0.29 | $0.84 | |
| $1.14 | $4.56 | N/A | — | — | — | — | — | — | |
| $0.32 | $0.89 | N/A | — | — | — | — | — | — | |
| $0.28 | $0.42 | $0.028 | — | — | — | — | — | — | |
| $0.27 | $1.10 | $0.07 | — | — | — | — | — | — | |
| $0.9 | $0.9 | N/A | — | — | — | — | — | — | |
| $0.2 | $0.2 | N/A | — | — | — | — | — | — | |
| $0.89 | $0.89 | N/A | — | — | — | — | — | — | |
| $0.32 | $0.89 | N/A | — | — | — | — | — | — | |
| $1.45 | $1.45 | N/A | — | — | — | — | — | — | |
| $1.25 | $1.25 | N/A | — | — | — | — | — | — | |
| $0.9 | $0.9 | N/A | — | — | — | — | — | — | |
Cost Calculator
US Dollar ($)
Preset:
Cheapest Instances to Run It
Cloud GPU instances that can host DeepSeek V3, ranked by cheapest on-demand price. The model needs about 1643 GB of GPU memory at FP16 precision (estimated from its parameter count), so treat the fit as guidance rather than a guarantee.
All clouds
FP16 (full precision)
US Dollar ($)
Instance | Cloud | GPU | VRAM | Price | Cheapest region |
|---|---|---|---|---|---|
| p6-b300.48xlarge |  AWS | 8× B300 | 2149 GB | $142.42/hr |
Versions
| Version | Released | Context | Input / 1M | Output / 1M | Status |
|---|---|---|---|---|---|
| DeepSeek V3.2 | 164K | $0.259 | $0.380 | Deprecated | |
| DeepSeek V3.2 Speciale | 164K | $0.580 | $1.68 | Available | |
| DeepSeek V3.2 Thinking | 128K | $0.620 | $1.85 | Available | |
| DeepSeek V3.2 Exp | 164K | $0.270 | $0.410 | Deprecated | |
| DeepSeek V3.2 Exp Reasoning | — | — | — | — | Available |
| DeepSeek V3.2 Exp | — | — | — | — | Available |
| DeepSeek V3.1 Terminus | 164K | $0.270 | $1.00 | Deprecated | |
| DeepSeek V3.1 | 164K | $0.250 | $0.950 | Deprecated | |
| DeepSeek V3.1 671B | — | 164K | — | — | Available |
| DeepSeek V3 | 164K | $0.200 | $0.200 | Current | |
| DeepSeek V3 324 | — | 164K | $0.200 | $0.400 | Deprecated |