tokevenDeepSeek V3
DeepSeek V3 is DeepSeek's balanced performance model, priced at $0.27/MTok input and $1.10/MTok output with a 164K context window. Prices are off-peak rates from api-docs.deepseek.com (read 2026-10-11). Peak-hour prices are 2x the listed input and output rates — Tokeven does not model time-of-day tiers; off-peak is used as the base. Cache hit: $0.07/MTok input. No batch discount published. Not yet benchmarked, so never recommended.
Input $/MTok
$0.27
Output $/MTok
$1.10
Context window
164K
Cache read discount
74% off
DeepSeek V3 pricing breakdown
| Rate | Input $/MTok | Output $/MTok |
|---|---|---|
| Standard | $0.27 | $1.10 |
| Cached read (74% off input) | $0.07 | - |
| Cached write (no surcharge) | $0.27 | - |
| Batch (0% off) | $0.27 | $1.10 |
All rates in USD per million tokens. Prices verified against provider documentation.
DeepSeek V3 task benchmarks
Confidence scores (0-100) and cross-provider rank for each task type. Rank 1 is best across all 88 models.
Not benchmarked. The third-party benchmark source used for confidence scores does not cover DeepSeek models yet. Scores will appear here once independent benchmarks are published.
DeepSeek V3 cost by use case
Estimated cost per API call using default token counts for each workload.
| Use case | Input tokens | Output tokens | Cost / call |
|---|---|---|---|
| Code Generation | 2,000 | 4,000 | $0.004940 |
| Unit Test Generation | 3,000 | 5,000 | $0.006310 |
| SQL Generation | 1,500 | 1,000 | $0.001505 |
| API Generation | 3,000 | 6,000 | $0.007410 |
| Code Refactoring | 4,000 | 4,000 | $0.005480 |
| Code Review | 5,000 | 2,000 | $0.003550 |
| PR Review | 8,000 | 3,000 | $0.005460 |
| Security Review | 6,000 | 2,500 | $0.004370 |
| Bug Detection | 5,000 | 2,000 | $0.003550 |
| Document Summarization | 10,000 | 1,000 | $0.003800 |
| PDF Summarization | 15,000 | 1,500 | $0.005700 |
| Meeting Notes | 8,000 | 2,000 | $0.004360 |
Monthly cost projections
Projected monthly spend at different call volumes, using code generation as a representative workload (2,000 input / 4,000 output tokens per call).
| Calls / month | Without caching | With caching (74% read discount) |
|---|---|---|
| 1K | $4.94 | $4.66 |
| 10K | $49.4 | $46.6 |
| 100K | $494 | $465.99 |
| 500K | $2,470 | $2,329.95 |
| 1M | $4,940 | $4,659.9 |
Caching estimate assumes 70% of input tokens are cache reads. Your actual cache hit rate will depend on prompt structure and reuse.
Mid-tier alternatives to DeepSeek V3
Same-tier models from other providers, sorted by input price.
| Model | Provider | Input $/MTok | Output $/MTok | Input savings |
|---|---|---|---|---|
| MiniMax M2 | MiniMax | $0.26 | $1.00 | 4% cheaper |
| Kimi K2 | Moonshot (Kimi) | $0.95 | $4.00 | 252% more |
| GPT 5 | GPT | $1.25 | $10.00 | 363% more |
| GPT 5.1 | GPT | $1.25 | $10.00 | 363% more |
| Gemini 2.5 Pro | Gemini | $1.25 | $10.00 | 363% more |
| Grok 4.3 | Grok | $1.25 | $2.50 | 363% more |
| Grok 4.20 Non-Reasoning | Grok | $1.25 | $2.50 | 363% more |
| Grok 4.20 Reasoning | Grok | $1.25 | $2.50 | 363% more |
| Grok 4.20 Multi-Agent | Grok | $1.25 | $2.50 | 363% more |
| GPT 5.2 | GPT | $1.75 | $14.00 | 548% more |
| Claude Sonnet 5 | Claude | $2.00 | $10.00 | 641% more |
| Claude Sonnet 5.5 | Claude | $2.00 | $10.00 | 641% more |
| GPT 5.6 Terra | GPT | $2.00 | $12.00 | 641% more |
| GPT 4.1 | GPT | $2.00 | $8.00 | 641% more |
| o3 | GPT | $2.00 | $8.00 | 641% more |
| Gemini 3.1 Pro | Gemini | $2.00 | $12.00 | 641% more |
| Gemini 3.1 Pro Custom Tools | Gemini | $2.00 | $12.00 | 641% more |
| GPT 5.4 | GPT | $2.50 | $15.00 | 826% more |
| GPT 4o | GPT | $2.50 | $10.00 | 826% more |
| Claude Sonnet 4.6 | Claude | $3.00 | $15.00 | 1011% more |
| Claude Sonnet 4.5 | Claude | $3.00 | $15.00 | 1011% more |
Track your DeepSeek V3 spend
See exactly how much you spend on DeepSeek V3 and where you can save. Two-line wrapper install, no API keys shared.