DeepSeek-V4-Pro vs DeepSeek-V3
Specs, VRAM requirements and download trends — updated 10 November 2026.
deepseek-ai · A · Open weights
Params
1,598.8 B
Context
1,048,576
30d
336 K
mit · commercial OK
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | DeepSeek-V4-Pro | DeepSeek-V3 |
|---|---|---|
| Deployment | Open weights · also via API | Open weights · also via API |
| Parameters | 1,598.8 B | 684.5 B |
| Architecture | DeepseekV4ForCausalLM | DeepseekV3ForCausalLM |
| Context length | 1,048,576 | 163,840 |
| Licence | mit | — |
| Commercial use | Allowed | Unknown |
| Languages | — | — |
| Downloads 30d | 335,647 | 1,487,156 |
| Downloads all time | 11.2 M | 23.5 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2026-04-22 | 2024-12-25 |
| API input price / 1M tokens | $0.96 | $0.32 |
| API output price / 1M tokens | $1.91 | $0.89 |
| Est. cost per 1M input + 1M output tokens | $2.87 | $1.21 |
| Cached input / 1M tokens | $0.08 | — |
| AA Intelligence Index | 12.6 | — |
| AA Coding Index | — | — |
| AA Agentic Index | — | — |
| Max output tokens | 384,000 | 16,384 |
| API providers | 12 | 2 |
| Tool calling | Yes | Yes |
| Reasoning mode | Yes | No |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
DeepSeek-V4-Pro
$2.87
DeepSeek-V3
$1.21
DeepSeek-V3 is 2.4× cheaper at list price.
Download trend
Daily snapshots, last 90 days (13 Aug – 10 Nov)
1.7 M334 K
DeepSeek-V4-Pro
DeepSeek-V3
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
DeepSeek-V3 · fp16 @ 160K ctx
2,811.5 GB / 24 GB
DeepSeek-V4-Pro · fp16 @ 1024K ctx
31,649.4 GB / 24 GB