DeepSeek Funding Docs Leaked: V4-Pro Matches Claude, Runs on Ascend
xiaohu · x · 2026-08-06
A document detailing DeepSeek's new funding round has leaked online, revealing key information about their next-generation model, V4-Pro.
Model Performance & Pricing
- V4-Pro ranks among the global top tier, with coding capabilities only 0.3% behind Anthropic's flagship Claude model.
- Supports 1 million token context length, with API pricing at 1/10th to 1/100th of overseas competitors, showing a massive lead in cost-effectiveness.
Compute & Ecosystem
- Natively adapted for Huawei Ascend 950PR chips, becoming a core pillar of China's three major independent 'model + chip' pathways and achieving a domestic compute closed-loop.
More from Venture
- DeepSeek's True Costs Revealed: Are Cheap Model APIs a Trap? — DanielLockyer · 2026-08-06
- Great Companies Are Built on Cheap Inputs: Anthropic and Scaling Laws — prestonpesek · 2026-08-06
- Report: DeepSeek Seeks $8B Funding at $74B Valuation — Polymarket · 2026-08-06
- Mirendil Inks $100M+ Google Cloud Deal to Scale Self-Improving AI — bneyshabur · 2026-08-06
- Indie Dev's $10k/100 Days Challenge: AI SEO Tool Hits $1,672 Revenue in 16 Days — ayushtweetshere · 2026-08-06
- Reality of AI Creator Economy: First Revenue Share Payout is Only $11.47 — Fun-Newspaper-83 · 2026-08-06