Deepseek V4 Pro

mit

DeepSeek-V4-Pro hybrid model with both non-thinking and thinking (default) modes.

Context
512K
Max output
384K
Input price
$0.15/1M tokens
Output price
$0.47/1M tokens

Capabilities

vision
tool call
structured output
reasoning
json mode
streaming
fine tuning
batch

Details

Model IDs/deepseek-v4-pro
Provider Together AI
Creator DeepSeek
Familydeepseek
Licensemit
Parameters—
Status—
Input modalitiestext
Output modalitiestext
Architecture—
Knowledge cutoff—
Training data cutoff—
Release date—
Deprecation date—
Typechat
Reasoning tokens—
Max input—
Open weightYes
Sourceofficial
Last updated

Pricing

Input
$0.15
↓93%$2
Output
$0.47
↓92%$6
Cache write
—
Cache read
—
Batch in
—
Batch out
—
Price history · $/1M input
$0.2
$0.2
$0.14
$3
$2
$0.15
Apr 25Jun 18Aug 31Sep 11Sep 13Sep 18
DateInputOutputChange
Apr 25, 2026$0.2$4.4—
Jun 18, 2026ab879c8$0.2$3.48↑0.0%
Aug 31, 2026e3d7ded$0.14$0.28↓30%
Sep 11, 202617bcfbf$3$15↑2043%
Sep 13, 2026f844c98$2$6↓33%
Sep 18, 20262ae168f$0.15$0.47↓93%

Input is ↓25% since Apr 25, 2026, from $0.2 to $0.15 per 1M tokens.

Family Comparison: deepseek

ModelContextMax outPricing
Deepseek V4 1 Flash——$0.01$1.2
Deepseek V4 Flash 0731——$0.03$0.28
Deepseek V4 Pro 0813512K—$0.13$3.96
Deepseek V4 Pro512K384K$0.15$0.47
Deepseek V4 Flash1M384K$0.3$1.2
Deepseek V3 2 Exp128K—$0.5$2.8
Deepseek V3 1128K—$0.6$1.7

API

GET/v1/models/together/s/deepseek-v4-pro

Deepseek V4 Pro

s/deepseek-v4-pro

DeepSeek-V4-Pro hybrid model with both non-thinking and thinking (default) modes.

Changes · 6 entries
s/deepseek-v4-proupdate2ae168fSep 18, 2026, 06:37 AM
pricinginput$2→$0.15output$6→$0.47
s/deepseek-v4-proupdatef844c98Sep 13, 2026, 06:38 AM
pricinginput$3→$2output$15→$6
s/deepseek-v4-proupdate17bcfbfSep 11, 2026, 06:36 AM
pricinginput$0.14→$3output$0.28→$15
s/deepseek-v4-proupdatee3d7dedAug 31, 2026, 04:13 AM
pricinginput$0.2→$0.14output$3.48→$0.28
s/deepseek-v4-proupdateab879c8Jun 18, 2026, 08:32 AM
pricingoutput$4.4→$3.48
s/deepseek-v4-procreatec8065d6Apr 25, 2026, 10:39 AM