DeepSeek V4 Pro — API Pricing & Benchmarks
DeepSeek V4 Pro is DeepSeek's flagship reasoning model, released in 2026-08. The API costs $1.32 per 1M input tokens and $3.96 per 1M output tokens ($1.98 blended at a 3:1 ratio), with cached input at $0.04. It supports a 1M-token context window. On the independently measured Artificial Analysis Intelligence Index it scores 53.2, ranking #22 of 68 models tracked on this site, with a median output speed of 67 tokens/sec. Pricing was changed on 2026-08-16 (previously $0.43/$0.87 per 1M tokens).
Explore provider routes and additional benchmark sources for DeepSeek V4 Pro
| Provider | DeepSeek |
| Input price / 1M tokens | $1.32 |
| Output price / 1M tokens | $3.96 |
| Cached input / 1M tokens | $0.04 |
| Blended price (3:1) | $1.98 |
| Context window | 1M |
| Max output tokens | 384K |
| Knowledge cutoff | — |
| License | open weights |
| Input modalities | T |
| Output speed (tok/s) | 67 |
| Released | 2026-08 |
| Artificial Analysis Intelligence Index | 53.2 |
| LMArena Elo | 1440 |
| GPQA Diamond | 92.8% |
| SWE-bench Verified | 96.4% |
| Humanity's Last Exam | 41% |
Where DeepSeek V4 Pro fits
On a blended 3:1 basis DeepSeek V4 Pro costs $1.98 per 1M tokens, cheaper than 48% of the 80 priced models in this index. As DeepSeek's flagship it is aimed at the hardest workloads — complex reasoning, agentic coding and multi-step tool use — where output quality dominates cost. Measured output speed of 67 tokens/sec is below the index median of 104 — typical of reasoning-heavy models, and worth checking against interactive latency budgets. The weights are open, so it can be self-hosted for data-residency or cost reasons instead of consumed via API. Cached input is priced at $0.04 (97% below standard input), which rewards prompt structures with long stable prefixes — see our caching guide.
Price history
| Period | Input $/1M | Output $/1M | Cached $/1M |
|---|---|---|---|
| 2026-07-18 → 2026-07-18 | $0.43 | $0.87 | $0.0036 |
| 2026-07-18 → 2026-08-16 | $0.43 | $0.87 | $0.0036 |
| 2026-08-16 → current | $1.32 | $3.96 | $0.04 |
List-price changes recorded by this site. The full dataset is public: price-history.json (CC BY 4.0).
Alternatives at a similar price
- Claude Haiku 4.5 (Anthropic) — $2.00/1M blended
- Muse Spark 1.3 (Meta) — $2.00/1M blended, AA Index 60.8
- Muse Spark 1.2 (Meta) — $2.00/1M blended, AA Index 56.8
- Muse Spark 1.1 (Meta) — $2.00/1M blended, AA Index 53.2
About DeepSeek
DeepSeek publishes open-weights models at disruptive prices. Even after its August 2026 switch to peak/off-peak billing, the V4 line costs a fraction of Western equivalents, and off-peak hours (everything outside 01:00–04:00 and 06:00–10:00 UTC on weekdays) halve the bill again.