DeepSeek Headlines

South China Morning Post

DeepSeek is recruiting about 150 senior backend engineers to overhaul infrastructure strained by surging user demand and compute-heavy AI agents, signaling its evolution into a high-volume platform.

Companies:DeepSeek

The Next Web

The Next Web covers DeepSeek's reported plan to order Huawei Ascend chips for a ~1GW data center, highlighting the strategic shift away from Nvidia as U.S. export restrictions persist.

Companies:DeepSeek

Bloomberg

Bloomberg reports that DeepSeek is planning a large order of Huawei AI chips for a new data center in Inner Mongolia, part of China's push to reduce dependence on Nvidia amid export restrictions.

Companies:DeepSeek

CNBC

A CNBC analysis of DeepSeek founder Liang Wenfeng's quant fund High-Flyer and how DeepSeek is now seeking outside capital amid growing funding needs.

Companies:DeepSeek

South China Morning Post

DeepSeek is close to closing a new funding round that would value the company at about $74 billion, raising roughly 50 billion yuan from investors including existing backers and new funds.

Companies:DeepSeek

South China Morning Post

Chinese open-weight models, led by DeepSeek, reached a record 62% share of token volume on Vercel's AI Gateway; Vercel CEO Guillermo Rauch highlighted the shift away from proprietary models.

Companies:DeepSeekVercel

The Next Web

DeepSeek launched its Harness agentic coding tool and raised API prices for V4 Pro and V4 Flash, with peak rates quadrupling.

Companies:DeepSeek

SiliconANGLE

DeepSeek debuted V4 Flash Vision Exp, an experimental multimodal model in its V4 series, available via its paid developer platform. The model improved on V4 Flash in image analysis and matched or beat Anthropic's Opus 4.8 on several visual benchmarks.

Companies:DeepSeek

DeepSeek API Docs

DeepSeek launched the experimental multimodal model V4-Flash-Vision-Exp on its API platform, providing image understanding alongside V4-Flash text capabilities, and also introduced a Files API.

Companies:DeepSeek

Bloomberg

DeepSeek released an experimental multimodal version of its V4 Flash model, claiming performance near Anthropic's Opus 4.8.

Companies:DeepSeek

InfoQ

DeepSeek released DeepSeek Harness (dsh), an open-source execution runtime under the MIT license for building AI agents, featuring a micro-kernel plugin architecture. The developer preview is built on Cordis and lets developers swap model endpoints and runtime components via configuration.

Companies:DeepSeek

Forbes

DeepSeek launched V4 Pro at prices up to 14 times higher than its budget offerings, marking a strategic shift. This opinion piece examines how the premium pricing and fundraising challenge DeepSeek's earlier narrative that frontier AI is becoming inexpensive.

Companies:DeepSeek

The Register

The Register covers DeepSeek's early open-source release of its agent harness, built around the concept that models, tools, sessions, and other components are all plugins.

Companies:DeepSeek

South China Morning Post

SCMP reports that DeepSeek launched a developer preview of Harness, a modular agent framework, signaling the company's strategic pivot toward autonomous, agentic AI.

Companies:DeepSeek

DeepSeek API Docs

DeepSeek announced general availability of V4-Pro with major agent upgrades, flexible reasoning effort, native OpenAI Responses API support, and new peak/off-peak API pricing.

Companies:DeepSeek

Quartz

DeepSeek officially released V4-Pro, its flagship model, on app, web, and API after months in preview, with enhanced agent capabilities and benchmark results. A price increase is scheduled for August 16.

Companies:DeepSeek

Nikkei Asia

DeepSeek released the official V4 Pro model with enhanced agentic capabilities at substantially higher prices, alongside a free harness tool similar to Claude Code.

Companies:DeepSeek

Bloomberg

DeepSeek is raising prices for its V4 models more than fourfold during peak hours ahead of a potential IPO, effective Aug. 16.

Companies:DeepSeek

VentureBeat

DeepSeek launched the official version of DeepSeek-V4-Pro, an updated flagship model focused heavily on agentic workloads, alongside DeepSeek Harness v0.1, a new open-source agent harness that gives developers an alternative to integrated coding-agent environments such as Anthropic’s Claude Code.

Companies:DeepSeek

The New Stack

DeepSeek open sourced its agent harness under MIT, an extensible runtime where the model adapter, tool registry and agent loop are all swappable plugins.

Companies:DeepSeek

Reuters

Chinese luggage ‌maker Anhui Korrun (300577.SZ), opens new tab said on Thursday a fund it invested in deployed 2.9 billion yuan for an indirect 0.8265% stake in DeepSeek, implying the private AI startup was valued at 350.88 billion yuan ($51.82 billion).

Companies:DeepSeek

TechCrunch

DeepSeek, the Chinese large language model developer, is preparing for a 2027 IPO debut, but it could come as early as the end of this year as it also looks to raise around $1.5 billion in new funds at about a $71 billion valuation, Bloomberg reports.

Companies:DeepSeek

VentureBeat

DeepSeek's recent decision to drastically cut pricing on its V4-Pro model by 75% should have been unequivocally good news for enterprise AI vendors and developers. Instead, many are discovering that cheaper models don’t automatically translate into healthier margins.

Companies:DeepSeek

Investing.com

DeepSeek announced plans to double its workforce after raising $7.4 billion in its first external funding round.

Companies:DeepSeek

CryptoBriefing

DeepSeek plans to at least double its headcount after raising $7.4 billion in its first external funding round.

Companies:DeepSeek