DeepSeek ships V4-Flash-Vision-Exp, a multimodal model at V4-Flash pricing
DeepSeek put V4-Flash-Vision-Exp live on its API — an experimental vision variant it says holds V4-Flash's text, agent, and reasoning quality while closing most of the multimodal agent gap to Opus-4.8. Images bill at up to 384 tokens each at standard V4-Flash rates, accepted as base64, external URLs, or free Files API references, so visual agent workflows cost roughly what text ones do. The Opus-4.8 comparison rests on a chart with no published numeric scores, leaving the pricing, not the benchmark, as the verified part of this release.
Source: api-docs.deepseek.com ↗
V4-Flash-Vision-Exp matches DeepSeek-V4-Flash on text capabilities—including agents, reasoning, and world knowledge.
DeepSeek
Why this matters
- → Multimodal agents now cost-competitive with text-only workflows at V4-Flash rates
- → Closes gap to Opus-4.8 on vision benchmarks without price premium
- → Files API enables reusable image uploads, reducing bandwidth costs in agent loops
Vision agent pricing flattens