Wondering what DeepSeek V4.1 Flash API pricing actually looks like now that the new model has landed? The quick version is just below, and then I’ll walk you through the full rate card, the peak-hour catch most people miss, and what the V4 Pro retirement means for your bill. I’ve been running DeepSeek models inside my own agent stacks all year, so this is the practical breakdown, not the press-release version.
- DeepSeek V4.1 Flash was released on 10 September 2026, per the official DeepSeek API changelog, and you call it with the model name
deepseek-flash. - Input tokens cost $0.15 per million on a cache miss off-peak ($0.30 peak) — and just $0.003 per million on a cache hit off-peak ($0.006 peak).
- Output tokens cost $0.60 per million off-peak and $1.20 peak, with off-peak rates running at half the peak price.
- Context length is listed at 1M tokens with a maximum of 384K output tokens.
- DeepSeek V4 Pro is being retired: from 14 September 2026, requests to
deepseek-v4-proare routed to V4.1 Flash and billed at the cheaper V4.1 Flash price.
DeepSeek V4.1 Flash API Pricing: The Full Rate Card
Everything in this section comes straight from the official DeepSeek API pricing page and changelog, fetched on 11 September 2026 — the day after the model dropped. DeepSeek V4.1 Flash was released on 10 September 2026 as the smallest model in DeepSeek’s new architecture family, with native multimodal visual understanding baked in. The API model name is simply deepseek-flash.
| Token type | Off-peak (per 1M) | Peak (per 1M) |
|---|---|---|
| Input — cache hit | $0.003 | $0.006 |
| Input — cache miss | $0.15 | $0.30 |
| Output | $0.60 | $1.20 |
Two other specs matter for cost planning: the model lists a 1M context length and a maximum of 384K output tokens. That cache-hit price is the sleeper here. At $0.003 per million tokens off-peak, repeated system prompts, agent scaffolding and long-running sessions get almost free input once the cache warms up. If you’re building agents that loop over the same instructions hundreds of times — which is exactly what I teach inside the AI Profit Boardroom — cache design is now worth more than model choice for your bill.
Peak vs Off-Peak: How the Discount Windows Work
DeepSeek keeps its dual-pricing system for V4.1 Flash: peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, and off-peak rates run at exactly half the peak price. For UK readers, those peak windows are 2am–5am and 7am–11am UK time during BST — which means most of the UK working afternoon and evening is off-peak.
Practically, that changes how you schedule batch work. Bulk content generation, embeddings-style sweeps, overnight agent runs — push them into off-peak windows and you halve the bill without touching a line of prompt code. I run my own DeepSeek off-peak pricing playbook this way, and the same scheduling logic carries straight over to V4.1 Flash.
What the V4 Pro Retirement Means for Your Bill
This is the part of the announcement most people skimmed past. According to the official changelog, DeepSeek V4 Pro — which only went GA on 13 August 2026 — is being retired, and after 14 September 2026 all requests to deepseek-v4-pro are routed to V4.1 Flash and billed at the V4.1 Flash price.
If you’ve got production code pointed at deepseek-v4-pro, three things follow. First, your bill drops automatically, because Flash rates are lower. Second, your outputs may change character, because you’re silently getting a different model — retest your prompts rather than assuming parity. Third, the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp also route to V4.1 Flash now, so the experimental vision variant I covered in my DeepSeek V4 Flash vision write-up has effectively graduated into the main model. Vision is no longer an experiment — it’s native in the standard endpoint.
DeepSeek’s stated reason for the fast retirement is blunt: their testing shows V4.1 Flash beating V4 Pro on performance, cost, speed and total time, so there’s no pricing tier left for Pro to occupy. A one-day-old budget model replacing last month’s flagship tells you everything about the pace this market is moving at.
How DeepSeek V4.1 Flash API Pricing Fits an AI SEO Workflow
Here’s where I put my SEO hat on. Cheap cached input plus a 1M context window is close to the ideal profile for programmatic SEO and agent-driven content work: you load your brand guidelines, internal linking rules and product data once, then hammer the cache. The workflows I run through Hermes Agent with DeepSeek and the DeepSeek harness agent are exactly this shape — long-lived sessions, heavy instruction reuse, bursts of output.
Run the maths on a typical brief-to-draft pipeline: a 50K-token cached context re-read 100 times off-peak costs about $0.015 in cache-hit input. The output side dominates, and even that is $0.60 per million off-peak. For agencies, the constraint has clearly shifted from API budget to editorial quality control — which is a strategy problem, not a tooling one. If you want help thinking that through for your own pipeline, book a free SEO strategy session and we’ll go through it together.
The Benchmarks Behind the Price
DeepSeek published a benchmark set alongside the release: GPQA Diamond 90.9, Codeforces rating 3471, Terminal-Bench 2.1 at 90.6, HLE with tools at 63.9, and DeepSWE v1.1 at 74.2.
That said, the pricing facts don’t depend on the benchmarks. Even if V4.1 Flash merely matches V4 Pro in practice, you’re getting it at Flash rates with native vision and a huge cache discount — that’s the real story for anyone buying tokens at volume.
The bottom line on DeepSeek V4.1 Flash API pricing
DeepSeek V4.1 Flash API pricing is aggressive even by DeepSeek’s standards: $0.15 per million input tokens off-peak (cache miss), $0.003 on cache hits, $0.60 per million output, all doubling in peak windows — with a 1M context and 384K max output. The forced V4 Pro migration on 14 September 2026 means many teams get a price cut without lifting a finger, but you should re-test prompts because the model underneath is changing. If you’re building content or agent pipelines, schedule heavy jobs off-peak and architect for cache hits — that’s where the real savings hide. And if you’d rather skip the trial-and-error, the AI Profit Boardroom has the working templates already.
FAQ: deepseek v4.1 flash api pricing
Is DeepSeek V4.1 Flash cheaper than V4 Pro?
Yes. DeepSeek’s changelog states V4.1 Flash beats V4 Pro on performance, cost, speed and total time, and from 14 September 2026 all deepseek-v4-pro requests are billed at the lower V4.1 Flash price anyway.
What are DeepSeek’s peak hours for API pricing?
Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday. Everything else is off-peak, at half the peak rate — a genuinely large saving for scheduled batch work.
What model name do I use to call DeepSeek V4.1 Flash?
Use deepseek-flash. The legacy identifiers deepseek-v4-flash and deepseek-v4-flash-vision-exp are temporarily accepted and routed to V4.1 Flash for compatibility.
Does DeepSeek V4.1 Flash support vision?
Yes. Per the official changelog, native multimodal visual understanding is built into the new architecture — it’s no longer a separate experimental vision model.
What is the context window for DeepSeek V4.1 Flash?
The official pricing page lists a 1M context length with a maximum output of 384K tokens.
When is DeepSeek V4 Pro retired?
After 14 September 2026, requests to deepseek-v4-pro are routed to V4.1 Flash and billed at V4.1 Flash rates, per the official changelog.
Related reading
- DeepSeek off-peak pricing: how the discount windows work
- DeepSeek V4 Flash vision: what the experimental model could do
- Running Hermes Agent on DeepSeek
Ready to actually use this? The cheapest tokens in the world don’t rank pages — systems do. Join 3,700+ members inside the AI Profit Boardroom for the live calls, daily tutorials and done-for-you templates, or book a free SEO strategy session and I’ll show you where DeepSeek V4.1 Flash fits in your stack.
About the author: Julian Goldie is an SEO agency owner with 394K+ YouTube subscribers, a 100% Upwork job-success score, 75K+ community members across his groups, 10+ years in SEO, and a best-selling SEO book. Catch the daily AI SEO experiments on YouTube, learn the systems inside the AI Profit Boardroom, or book a free SEO strategy session. For agency work, book a call for a custom quote.
Last updated September 2026. This is the living guide to deepseek v4.1 flash api pricing — it gets updated as the tools change.