DeepSeek makes price cuts permanent - Asian Tech Roundup

Plus: SK Hynix hits market cap of $1 trillion

Image:
DeepSeek makes price cuts permanent

Welcome to Computing's weekly roundup of tech news in Asia. This time we look at the AI price war, SK Hynix joining the trillion-dollar club and a UK-Australia security collaboration agreement.

DeepSeek has made a 75% price cut on its V4‑Pro model permanent, rather than letting a promotional discount expire at the end of May.

China’s best-known AI startup will be keeping prices low for the foreseeable future, with typical API pricing around US $0.435 per million input tokens, making it more than 20 times cheaper than equivalent models from Google, OpenAI and Anthropic. At the same time, cache-hit pricing - crucial for repeated prompts and long-running AI agents - has been cut even further, down to one-tenth of previous costs.

This marks a further escalation in the global AI price war, at a time when many CIOs are questioning their spending on tokens. For high-volume users of coding models, AI agents, and other compute-intensive workloads, switching to DeepSeek could potentially save millions.

Rather than being merely a marketing push or a move to grab market share (although it is certainly those as well), DeepSeek is able to make these price cuts because V4-Pro has been engineered to drastically reduce its compute and memory footprint, according to analyst Sanchit Vir Gogia of Greyhound Research.

While many users in the West may be hesitant to move to Chinese alternatives due to concerns about data security and compliance - and while the likes of Gemini and Claude still have the edge in terms of benchmarks - DeepSeek’s decision to entrench lower prices at a time when many AI companies are looking increasingly financially unsustainable certainly turns up the pressure.

Australia

China

India

Japan

South Korea

Taiwan

Elsewhere in Asia