# DeepSeek V4.1 Flash Is Arriving - Try It Now on OpenCheese

> DeepSeek is preparing to officially release V4.1 Flash. According to the company’s testing, the new model outperforms V4 Pro across key metrics - including performance, speed, cost, and task completion time. OpenCheese users can try it right now.
>
> OpenCheese AI · 2026.09.10 · Web page: https://opencheese.dev/blog/deepseek-v4-1-flash-is-arriving-try-it-now-on-opencheese

DeepSeek plans to officially release **V4.1 Flash around September 10, 2026 (Beijing time)**.

According to DeepSeek, extensive internal and external testing shows that V4.1 Flash has surpassed V4 Pro across all key metrics, including overall performance, operating cost, response speed, and task completion time.

Following the official launch of V4.1 Flash—and until V4.1 Pro becomes available—all requests sent to the Pro model will be automatically routed to V4.1 Flash and billed at the lower Flash pricing.

## New DeepSeek V4.1 Flash pricing

The new pricing takes effect at **04:00 UTC on September 10, 2026**.

| Usage                        | Off-peak price | Peak-hour price |
| ---------------------------- | -------------: | --------------: |
| 1M input tokens — cache hit  |         $0.003 |          $0.006 |
| 1M input tokens — cache miss |          $0.15 |           $0.30 |
| 1M output tokens             |          $0.60 |           $1.20 |

**Peak hours in UTC:** 01:00-04:00 and 06:00-10:00, Monday through Friday. All other times are billed at off-peak rates.

V4.1 Flash promises a stronger balance between quality, speed, and price, making it an interesting option for agents, development workflows, automation, and high-volume tasks.

**You can try DeepSeek V4.1 Flash at [opencheese.dev](https://opencheese.dev/en/) right now!**

## Related posts

- 2026.09.29 · [GPT-6.1 Sol vs Astra and Claude: A Clear Comparison of Performance and Cost](https://opencheese.dev/blog/gpt-6-1-sol-vs-astra-and-claude-a-clear-comparison-of-performance-and-cost.md): GPT-6.1 Sol improves on GPT-6 Sol and comes close to Astra in independent testing. Here is how its benchmark results, task costs and token prices compare with competing models.
- 2026.10.31 · [DeepSeek and Moonshot reportedly trained new models on GPT-6 Astra outputs](https://opencheese.dev/blog/deepseek-moonshot-reportedly-trained-on-gpt-6-astra-outputs.md): A report citing unnamed sources says DeepSeek and Moonshot AI used distillation on GPT-6 Astra to build Kimi K3.1 Pro and DeepSeek V4.1-Flash. Neither company's benchmark claims have been independently confirmed.
- 2026.10.24 · [DHS bans DeepSeek across federal agencies, including systems already installed](https://opencheese.dev/blog/dhs-bans-deepseek-across-federal-agencies.md): The US Department of Homeland Security cites the risk of Chinese authorities reaching federal employees' data. The ban covers new purchases, existing installations and third-party products built on DeepSeek, with exceptions available on request.