DeepSeek-V4-Flash Tested: $0.31 for Tasks That Cost $35 on Other Models
Teknium · x · 2026-08-05
A developer reported running DeepSeek-V4-Flash 0731 through multi-hour, challenging tasks for only $0.31. In contrast, completing similar tasks last week with Kimi K3 and similarly priced models cost $35.
The author teased an upcoming YouTube video showcasing the outputs, noting that the expensive models were not 100x better.
More from Models
- DeepSeek V4 Flash Matches GLM 5.2, Tops OpenRouter Rankings — natolambert · 2026-08-05
- DeepSeek V4 Flash Surfaces on OpenRouter: 284B Total Params, 1M Context — MikePFrank · 2026-08-05
- Netizen Tests Minimax H3: Generates Flawless Code with a Single Prompt — CSProfKGD · 2026-08-05
- New Claude Models Excel at Complex Tasks but Struggle with Everyday Clarity — generativist · 2026-08-05
- Token Prices in Historic Freefall as Compute Shifts to Cheaper Chinese AI Models — ccerrato147 · 2026-08-05
- Leaked SpaceX/xAI Q2 Updates: Grok to Ingest All SpaceX Data, Targets $100B ARR by Year-End — ns123abc · 2026-08-05