Briefing · new model
DeepSeek releases V4.1-Flash with new pricing effective Sept 10, 2026
DeepSeek has released DeepSeek-V4.1-Flash, a 552B-parameter MoE model with native visual understanding. The release includes new API pricing effective September 10, 2026, where off-peak rates are 50% of peak rates. The post also announced that deepseek-v4-pro requests would be routed to V4.1-Flash from September 14, 2026; DeepSeek later said it would keep providing V4 Pro on its API after that date. Source: DeepSeek, read on .
What the source says
The new model features a 552B-parameter Mixture-of-Experts architecture with specific active parameter counts.
“🔹 552B-parameter MoE. 🔹 New Causal Encoder–Decoder architecture: just 8B active parameters for input, 16B for output.”
New API pricing takes effect on September 10, 2026, with distinct peak and off-peak rates.
“🔹 New pricing takes effect at 04:00 UTC on Sept 10, 2026.”
Off-peak rates are set at half the cost of peak rates.
“🔹 Off-peak rates are 50% of peak rates.”
Later update from DeepSeek
The source also said:
“Starting at 04:00 UTC on Sept 14, 2026, all deepseek-v4-pro requests will route to V4.1-Flash at V4.1-Flash rates.”
DeepSeek API change log, in its entry dated , says:
“In response to user demand, we have decided to continue providing API services for DeepSeek V4 Pro after September 14, 2026, with the billing method remaining unchanged.”
Source: DeepSeek, read on .
Where this lands in our prices Calculated cost per task at today's listed prices
Models named in the source that we already track, with the standard tasks recalculated by our engine at the prices listed today — visible output only, no cache discount. These numbers are ours, not the source's.
| Model | L1 micro | L2 analysis | L3 agentic | |
|---|---|---|---|---|
| DeepSeek: DeepSeek V4.1 Flash (batch) | $0.0001 | $0.0023 | $0.016 | Open in the calculator |
| DeepSeek: DeepSeek V4 Flash 0423 | $0.0002 | $0.015 | $0.121 | Open in the calculator |
| DeepSeek: DeepSeek V4 Flash Vision Exp | $0.0009 | $0.074 | $0.591 | Open in the calculator |
What to check
- Update API calls to use 'deepseek-flash' instead of retired models.
- Review cost estimates against new peak/off-peak pricing structures effective Sept 10, 2026.
- Verify infrastructure requirements for reduced KV cache storage needs (1/4 HBM, 1/8 SSD).
- The dates in the source — , — had already passed when we read it on : check DeepSeek’s current documentation for what applies now.
How this page is made
The quotes above are copied word for word from the source page as read on 17 Oct 2026 and checked against its text. Any task-cost figures are calculated from our own tracked prices on the build date; the source did not provide them. This is a dated record of what the source published, not advice.