DeepSeek V4 Arrives With Near State-of-the-Art Intelligence At 1/6th the Cost

Source: Slashdot

Introduction

An anonymous reader quotes a report from VentureBeat: The whale has resurfaced. DeepSeek, the Chinese AI startup spun out of High‑Flyer Capital Management’s quantitative analysis firm, became a near‑overnight sensation globally in January 2025 with the release of its open‑source R1 model that matched proprietary U.S. giants. It marked an epoch in AI, and while DeepSeek has released several updates to that model and its other V3 series, the international AI and business community has been waiting with baited breath for the follow‑up to the R1 moment.

DeepSeek‑V4 Release

Last night DeepSeek announced DeepSeek‑V4, a 1.6‑trillion‑parameter Mixture‑of‑Experts (MoE) model available for free under a commercially‑friendly open‑source MIT License. On several benchmarks it nears—or even surpasses—the performance of the world’s most advanced closed‑source systems, while costing roughly 1/6th as much to use via an API.

The release was described by DeepSeek AI researcher Deli Chen on X as a “labor of love” 484 days after the launch of V3, and he noted, “AGI belongs to everyone.” The model is now available on the AI code‑sharing community Hugging Face and through DeepSeek’s own API.

Pricing and Economic Impact

The DeepSeek‑V4‑Pro variant delivers “near‑frontier performance” at a substantially lower price point:

• $5.22 for 1 million input tokens and 1 million output tokens
• Compared with $35 for GPT‑5.5 and $30 for Claude Opus 4.7

This translates to roughly 1/7th the cost of GPT‑5.5 and 1/6th the cost of Claude Opus 4.7, supporting VentureBeat’s observation that DeepSeek is “compressing advanced model economics into a much lower band.”

Outlook

While GPT‑5.5 and Claude Opus 4.7 still lead on most benchmarks, DeepSeek‑V4‑Pro’s competitive performance combined with its lower cost could “force a major rethink of the economics of advanced AI deployment.”

Read more of this story at Slashdot.