GPT-56T 861 —
MUSE-SPK 835 -0.7%
GPT-56SC 828 -5.2%
QWEN-38X 824 —
CL-OP55X 822 —
GROK-46H 822 -5%
GPT-6A 820 —
GLM-5 784 -8.4%
CL-FAB5H 743 -5.6%
KIMI-K3X 742 -8.4%
CL-OP5H 720 -5.8%
CL-OP5X 709 -18%
CL-OP46H 698 -5.9%
CL-OP47H 690 -5.9%
GEM-38FH 677 +0.1%
GEM-37FH 657 -24%
GPT-56S 622 —
CL-OP47 582 -0.7%
GPT-55H 582 —
INKL 531 —
GEM-31P 513 —
GEM-3P 499 —
CL-OP46 496 -0.2%
CL-OP48 490 —
GPT-56T 861 —
MUSE-SPK 835 -0.7%
GPT-56SC 828 -5.2%
QWEN-38X 824 —
CL-OP55X 822 —
GROK-46H 822 -5%
GPT-6A 820 —
GLM-5 784 -8.4%
CL-FAB5H 743 -5.6%
KIMI-K3X 742 -8.4%
CL-OP5H 720 -5.8%
CL-OP5X 709 -18%
CL-OP46H 698 -5.9%
CL-OP47H 690 -5.9%
GEM-38FH 677 +0.1%
GEM-37FH 657 -24%
GPT-56S 622 —
CL-OP47 582 -0.7%
GPT-55H 582 —
INKL 531 —
GEM-31P 513 —
GEM-3P 499 —
CL-OP46 496 -0.2%
CL-OP48 490 —
← Back to feed

DeepSeek Locks In V4 Pro's 75% Price Cut as Permanent — $0.87/M Output From June 1

DeepSeek has quietly confirmed in its official API documentation that the 75% promotional price cut applied to V4 Pro is not a temporary discount — it becomes the permanent rate when the promotion window closes on May 31, 2026 at 15:59 UTC.

The official API docs state: “The deepseek-v4-pro model API pricing will be officially adjusted to 1/4 of the original price after the 75% discount promotion ends on 2026/05/31 15:59 UTC.” That means the promotional and permanent prices are identical: the discount absorbs into the list price.

What the Numbers Look Like

ModelInput (cache miss)OutputContext
DeepSeek V4 Pro (permanent from June 1)$0.435/M$0.87/M1M
DeepSeek V4 Flash$0.14/M$0.28/M1M
GPT-5.5$5/M$30/M128K
Claude Opus 4.7$15/M$75/M200K
Gemini 3.1 Pro Preview$1.25/M$5/M1M

At $0.87/M output, V4 Pro sits 34x below GPT-5.5 and 86x below Opus 4.7. Gemini 3.1 Pro Preview is the closest frontier comparison at $5/M output — still nearly 6x more expensive.

Cache hit pricing was separately cut to one-tenth of the original rate on April 26 and remains in effect: $0.0145/M input (cache hit) vs $0.435/M (cache miss), and $0.003625/M for cache hits on Pro.

Performance Context

This is not a cheap model with cheap capability. V4 Pro launched April 24, 2026 as an open-weight 1.6T parameter MoE with 49B active parameters and a genuine 1M context window. On the MRCR benchmark at 1M context, V4 Pro posts 83.5% recall — above GPT-5.5 (74%) and well above Opus 4.7 (32%) at that length.

SWE-bench Verified puts V4 Pro at 80.6%, matching Gemini 3.1 Pro Preview and near the frontier ceiling.

The API is also Anthropic-format compatible. Migrating requires changing one URL and one model string in any Anthropic SDK — no rewrite, no new tooling.

Why This Matters

The V4 Flash model has always been the cost-optimised tier at $0.28/M output. V4 Pro now sits at a price point that was, six months ago, mid-tier territory. The gap between frontier capability and budget deployment is collapsing.

For developers running agentic workloads at scale, the calculus is shifting: a 1M context model with strong agentic performance at under $1/M output is a different category of tool than it was at $3.48/M. DeepSeek is making V4 Pro the default serious-but-affordable option rather than the budget-compromise one.

The permanent pricing announcement also arrives as DeepSeek closes its first external funding round — reportedly approaching $10B — meaning price cuts are now underpinned by fresh external capital rather than solely founder subsidy.