DeepSeek-V3.2 vs DeepSeek-V4-Flash-0731
Specs, VRAM requirements and download trends — updated 24 October 2026.
deepseek-ai · A · Open weights
Params
685.4 B
Context
163,840
30d
5.4 M
mit · commercial OK
deepseek-ai · B · Open weights
Params
304.2 B
Context
1,048,576
30d
4.3 M
mit · commercial OK
Specification comparison
Differences are highlighted; identical values are muted.
| Specification | DeepSeek-V3.2 | DeepSeek-V4-Flash-0731 |
|---|---|---|
| Deployment | Open weights · also via API | Open weights · also via API |
| Parameters | 685.4 B | 304.2 B |
| Architecture | DeepseekV32ForCausalLM | DeepseekV4ForCausalLM |
| Context length | 163,840 | 1,048,576 |
| Licence | mit | mit |
| Commercial use | Allowed | Allowed |
| Languages | — | — |
| Downloads 30d | 5,435,964 | 4,348,313 |
| Downloads all time | 30.4 M | 11.9 M |
| Quantizations on the Hub | SAFETENSORS | SAFETENSORS |
| Gated | No | No |
| First seen on the Hub | 2025-12-01 | 2026-07-31 |
| API input price / 1M tokens | $0.28 | $0.044 |
| API output price / 1M tokens | $0.42 | $0.13 |
| Est. cost per 1M input + 1M output tokens | $0.70 | $0.18 |
| Cached input / 1M tokens | $0.028 | $0.0014 |
| AA Intelligence Index | 21.5 | 34.3 |
| AA Coding Index | 44.2 | 69.1 |
| AA Agentic Index | — | 41 |
| Max output tokens | 65,536 | 131,072 |
| API providers | 11 | 22 |
| Tool calling | Yes | Yes |
| Reasoning mode | Yes | Yes |
API cost
Estimated cost of 1M input + 1M output tokens via OpenRouter
DeepSeek-V3.2
$0.70
DeepSeek-V4-Flash-0731
$0.18
DeepSeek-V4-Flash-0731 is 3.9× cheaper at list price.
Download trend
Daily snapshots, last 89 days (28 Jul – 24 Oct)
5.5 M236 K
DeepSeek-V3.2
DeepSeek-V4-Flash-0731
A overtook B on 11 Oct
VRAM side by side
On RTX 4090 · 24 GB · 8K context unless noted
DeepSeek-V3.2 · fp16 @ 160K ctx
2,815.1 GB / 24 GB
DeepSeek-V4-Flash-0731 · fp16 @ 1024K ctx
6,024.3 GB / 24 GB