OpenAI Cuts Prices to Defend, DeepSeek Raises to Capture Value
创业邦 · wechat · 2026-08-19
In the summer of 2026, the global AI industry witnessed a "reverse operation": OpenAI slashed the price of its lightweight model Luna by 80%, while DeepSeek announced a significant price hike. This reflects divergent strategic paths between Chinese and American AI companies.
OpenAI: Layered Defense and Volume Capture
OpenAI built a three-tier matrix around GPT-5.6: Sol (Flagship), Terra (Mid-range), and Luna (Lightweight). Luna's price drop to $0.2 (input) aims to penetrate the low-end market and defend against cost-effective Chinese models. Sol maintains high pricing and offers a Fast acceleration mode, securing profits for high-value scenarios. The strategy shifts from "flagship premium" to "layered routing driving total volume".
DeepSeek: Parity and Commercial Correction
DeepSeek's confidence to raise prices rests on:
- Capability Parity: V4-Flash's Intelligence Index is only 1 point lower than GPT-5.6 Luna.
- Cost Advantage: Via CSA+HCA hybrid attention and sparse selection, its per-token cost is a fraction of overseas models.
Morgan Stanley interprets this as a shift from price wars to "commercialization driven by intelligent capability".
Industry Landscape: The Rise of the East
- Pricing Power Shift: Chinese models are becoming the new pricing benchmark.
- Market Segmentation: High-cost-of-failure markets (complex agents) remain dominated by GPT-5.6 Sol and Claude; high-volume automation markets (summarization, extraction) become the battlefield for cost-effectiveness.
- Data Evidence: DeepSeek's token usage share among US enterprises has doubled, with significant gains on the OpenRouter platform, occasionally surpassing US models.
More from Companies & People
- MIT PhD Sharut Gupta joins DeepMind; Isola hails her representational alignment defense talk — phillip_isola · 2026-10-16
- Google DeepMind is hiring a 19-week Student Researcher on scaling laws in San Francisco — savvyRL · 2026-10-16
- 21-person team ships hundreds of PRs weekly with AI writing 97% of the code — rseroter · 2026-10-16
- Cursor's final interview: a desk, a laptop, two days to build anything — a16z · 2026-10-16
- Dev workshops: having attendees build what you didn't teach is the real test — gethackteam · 2026-10-16
- 80% of enterprises this consultant works with run on Databricks, and can't leave — verrsane · 2026-10-16