DeepSeek V4 Flash Vision Exp API Prices Double Across Input and Output

The Numbers

DeepSeek has doubled the API pricing on V4 Flash Vision Exp, effective now. Input tokens move from $0.22 to $0.44 per 1M tokens (+100%), and output tokens jump from $0.66 to $1.32 per 1M tokens — also a clean +100% increase across the board.

No partial adjustments, no tiered rollout. Both sides of the token ledger doubled simultaneously.

Does This Matter for Your Workload?

It depends on your volume and output-heaviness. At the old rates, a workload generating 10M output tokens per month cost $6.60. At the new rates, that same workload costs $13.20 — a meaningful line item if you're running vision tasks at scale.

For lighter users doing occasional image analysis or prototyping, the absolute dollar impact stays small. At $0.44 input and $1.32 output, this model is still in the budget tier relative to many frontier vision models — but the gap just narrowed.

Output-heavy pipelines (long descriptions, OCR-heavy workflows, detailed captioning) will feel this most. Input-heavy workloads with short responses get hit less in absolute terms, though the 100% input increase still stings on large batch jobs.

What to Do Next

If V4 Flash Vision Exp is a significant cost driver, now is a good time to benchmark alternatives. Check the DeepSeek V4 Flash Vision Exp — live specs & price history page for up-to-date positioning against comparable vision models.

If your use case is tolerant of slightly lower throughput, a heavier model at a competitive price point may now offer better value-per-output than it did before this increase.

Ezra, Scout AI Team