Models38in catalog
Largest context1.0MDeepSeek V4 Flash

Avg context

166K

Cheapest input

Free/M

Deepseek V3 2 251201

Largest output

384K

DeepSeek V4 Flash

Speed tiers

fast 8balanced 25deep 5

Capabilities

What Deepseek models support across the catalog.

Vision

1 / 38

3% of models

Tool use

21 / 38

55% of models

Function calling

21 / 38

55% of models

Extended thinking

22 / 38

58% of models

Streaming

38 / 38

100% of models

Prompt caching

11 / 38

29% of models

All Deepseek models

Sorted by context window, largest first. Tap a row for full specs.

ModelContext
DeepSeek V4 Flash1.0M
DeepSeek V4 Pro1.0M
Deepseek Coder V2 Lite164K
Deepseek Coder V2 Lite Base164K
Deepseek Prover164K
Deepseek V2 Lite164K
DeepSeek V3164K
Deepseek V3 1164K
DeepSeek V3.1164K
Deepseek V3 1 Maas164K
Deepseek V3 1 Terminus164K
DeepSeek V3.2164K
DeepSeek V3.2 Exp164K
Deepseek V3 2 Maas164K
DeepSeek V3.2 Speciale164K
Deepseek V3p2164K
R1 0528164K
Deepseek Prover V2 671b160K
Deepseek R1 671b131K
Deepseek Reasoner131K
Deepseek128K
Deepseek Coder128K
Deepseek R1 0528 Tput128K
Deepseek R1 Basic128K
Deepseek V3 0324128K

Showing 25 of 38 models

See rankings

How Deepseek models rank against the full catalog.