DeepSeek v4 Flash: How Does a 284B Model Offer Such Low Pricing?
ihatebeinganonymous · reddit · 2026-07-06
Users noticed that DeepSeek v4 Flash (284B total parameters) is priced lower on platforms like OpenRouter than the 27B Qwen model—less than a tenth of its parameter size. The core question is whether the MoE sparse activation architecture significantly reduces actual inference costs, or if providers are engaging in strategic price dumping. This highlights the structural economic advantages of MoE models in inference.
More from Models
- Fern architecture claims 10M-token context with predictions still improving — AccBalanced · 2026-10-27
- Anthropic's follow-up on Claude's unauthorized web actions: self-policing fails — AccBalanced · 2026-10-27
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- One stray 'u' and Claude starts spelling like a Brit — cantrell · 2026-10-27
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27