DeepSeek V4 Pro repriced across 3 hosts: Sail Research (US region) input down 47%, CoreWeave output up 55%

Published 17 September 2026

Price move, per 1M tokens

CoreWeave · input↑ 14%
$1.15 → $1.31
CoreWeave · output↑ 55%
$2.55 → $3.96
CoreWeave · cache read↓ 78%
$0.200 → $0.044
Sail Research (US region) · input↓ 47%
$1.31 → $0.70
Sail Research (US region) · output↓ 25%
$3.94 → $2.96
Sail Research (US region) · cache read↓ 25%
$0.044 → $0.033
Sail Research · input↓ 47%
$1.31 → $0.70
Sail Research · output↓ 25%
$3.94 → $2.96
Sail Research · cache read↓ 25%
$0.044 → $0.033
Ionstream · input↓ 8%
$0.96 → $0.88
Ionstream · outputunchanged
$2.88 → $2.88
Ionstream · cache readunchanged
$0.088 → $0.088

host	rate	was	now	change
CoreWeave	input	$1.15	$1.31	↑ 14%
CoreWeave	output	$2.55	$3.96	↑ 55%
CoreWeave	cache read	$0.200	$0.044	↓ 78%
Sail Research (US region)	input	$1.31	$0.70	↓ 47%
Sail Research (US region)	output	$3.94	$2.96	↓ 25%
Sail Research (US region)	cache read	$0.044	$0.033	↓ 25%
Sail Research	input	$1.31	$0.70	↓ 47%
Sail Research	output	$3.94	$2.96	↓ 25%
Sail Research	cache read	$0.044	$0.033	↓ 25%
Ionstream	input	$0.96	$0.88	↓ 8%
Ionstream	output	$2.88	$2.88	—
Ionstream	cache read	$0.088	$0.088	—

CoreWeave output: ↑ 55%CoreWeave cache read: ↓ 78%Sail Research (US region) input: ↓ 47%Sail Research input: ↓ 47%Ionstream input: ↓ 8%

The model

DeepSeek V4 Proopen the model →

Open weights1598.8B parameters1049K context28 hosts

intelligence36th of 168 writing32nd of 168 coding43rd of 168 agents31st of 55

DeepSeek V4 Pro is a large downloadable text model with a one-million-token request limit and strong mathematics scores. Its mixture-of-experts design keeps only 37 billion parameters active per token, making it more efficient to run than its total size suggests.

Source

Read from the hosts' own published rates · we published this on 17 September 2026.

← all news