DeepSeek V4 Flash Vision Exp API Prices Double: +104% on Input and Output

The Numbers

DeepSeek has quietly doubled the cost of its V4 Flash Vision Exp API. Input tokens moved from $0.216 to $0.44 per 1M tokens, and output tokens jumped from $0.647 to $1.32 per 1M tokens — both changes landing at exactly +104%.

That's not a rounding adjustment. That's a full reprice.

Does This Matter for Your Workload?

At the old rates, this model was priced aggressively — likely a launch-period discount. The new numbers are still competitive in the multimodal vision space, but the math changes meaningfully at scale.

A workload burning 100M output tokens per month goes from roughly $64.70 to $132.00 — a delta of ~$67 monthly on output alone. Teams running high-volume document parsing, image captioning pipelines, or vision-augmented RAG systems should rerun their cost projections now.

Lower-volume users (under ~10M tokens/month) will feel this less acutely — we're talking single-digit dollar differences — but it's still worth auditing if you locked in budget assumptions at the old price.

What To Do

• Re-benchmark your pipeline. If you selected this model partly on price, the cost-to-quality calculus has shifted.
• Check alternatives. Other multimodal models on our site may now be comparably priced or cheaper depending on your input/output ratio.
• Watch for further changes. A 104% single move suggests initial pricing was provisional. More adjustments are possible.

See current rates and track future changes on the DeepSeek V4 Flash Vision Exp — live specs & price history page.

Ezra, Scout AI Team