DeepSeek V4 Pro, V4 Flash, R1 Reasoning, and V3 Chat: 1M token context, disruptive low pricing, peak/off-peak tier discounts, and 90% cache-hit savings.
7 models tracked · cheapest input: DeepSeek V3 (Chat) at $0.14/M · official pricing page
DeepSeek V4 delivers flagship-adjacent output at a fraction of Western pricing — off-peak hours cut rates a further 50% and cache-hit input is ~97% cheaper. It is the strongest budget choice for high-volume, price-sensitive workloads.
deepseek-v4-pro
Input /M
$1.32
Output /M
$3.96
Context
1M
Next-gen MoE frontier model with 1M context, 384k max output, and dynamic off-peak discounts ($0.66/$1.98 off-peak, $1.32/$3.96 peak).
deepseek-v4-flash
Input /M
$0.44
Output /M
$1.32
Context
1M
Ultra-cheap 1M context model at $0.44/$1.32 per 1M tokens ($0.22/$0.66 off-peak) with cache hit rates down to $0.007.
deepseek-reasoner
Input /M
$0.55
Output /M
$2.19
Context
64K
Open-weights reasoning breakthrough: rivals OpenAI o1 on math, coding, and logical reasoning at a fraction of the cost ($0.55/$2.19).
deepseek-chat
Input /M
$0.14
Output /M
$0.28
Context
64K
671B parameter MoE architecture (37B active) delivering frontier chat performance at $0.14/$0.28 per 1M tokens.
deepseek-coder
Input /M
$0.14
Output /M
$0.28
Context
128K
236B MoE coding model supporting 338 programming languages with fill-in-the-middle capability and 128k context.
deepseek-v3.2
Input /M
$0.26
Output /M
$0.38
Context
163.8K
DeepSeek reasoning model route with integrated thinking for tool use and a 163K context window.
deepseek-v4-flash-vision-exp
Input /M
$0.44
Output /M
$1.32
Context
1.1M
Experimental vision-enabled DeepSeek V4 Flash (284B MoE, 13B active): adds image understanding to the V4 Flash base with the same text capabilities, agents and reasoning.
Other providers
All DeepSeek prices verified Aug 28, 2026. Prices change frequently — each model page links to the authoritative source.