DeepSeek · llm model
DeepSeek-V4.1-Flash-1017
DeepSeek-V4.1-Flash-1017 is DeepSeek's latest language model in our dataset, released Oct 17, 2026. Open weights (MIT). DeepSeek's Oct 17, 2026 update to V4.1 Flash: MIT-licensed open weights, now served behind the deepseek-flash API name, with shorter outputs and faster task completion than the Sep 10 version. Verified against vendor sources on Oct 26, 2026.
Reported Last verified · All models · JSON · Markdown
- ReleasedOct 17, 2026
- Context window1M1,000,000 tokens
- Max output384K384,000 tokens
- LicenceOpen weightsMIT
Key facts
| Fact | Value | Source |
|---|---|---|
| Vendor | DeepSeek | n/a |
| Released | Oct 17, 2026 DeepSeek's release post and changelog entry are dated 2026/10/17. | source |
| Status | generally available | n/a |
| API id | deepseek-flash The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are routed to this model. | source |
| Context window | 1,000,000 tokens (1M) | source |
| Max output | 384,000 tokens | source |
| Input | text, image Native image understanding per DeepSeek's pricing page. | source |
| Output | text | source |
| Licence | MIT (open weights) License as listed on the Hugging Face model repository. | source |
| Pricing | ¥1 in / ¥4 out per 1M tokens; peak: ¥2 / ¥8 Official price list is in CNY. Off-peak shown; peak (Beijing time Mon-Fri 09-12 and 14-18) is double. DeepSeek's Oct 17 release post lists no price change. We do not convert currencies. | source |
Where DeepSeek-V4.1-Flash-1017 sits in our data
- Released Oct 17, 2026 (week 42 of 2026), in October 2026. Other models in our data released that month: GPT-6.1 Terra, Kimi K3.1 Pro, Kimi K3.1 Pro Highspeed, Nemotron 3.5 Bolt E4B (4GB), Voxtral 3, Mercury 3.
- The previous DeepSeek model we track is DeepSeek-V4.1-Flash (Sep 10, 2026), 37 days earlier.
- Of the 60 models in our data with a published context window, DeepSeek-V4.1-Flash-1017 has the number 22 largest (1M tokens). Our data is not every model on the market.
- DeepSeek tools we cover: DeepSeek.
- All releases by date: AI model releases by month · new AI models this week. Choosing for code: best LLM for coding.
How DeepSeek-V4.1-Flash-1017 differs from DeepSeek-V4.1-Flash
Read from the recorded facts of the two models, not from a head-to-head test. We have not run either.
- Listed price is the same as DeepSeek-V4.1-Flash: CNY 1 in and CNY 4 out (per 1M tokens).
- Both list a 1M-token context window.
- Both accept text, image as input.
- DeepSeek-V4.1-Flash-1017 came out 37 days after DeepSeek-V4.1-Flash.
- Facts for the other model: DeepSeek-V4.1-Flash.
What the sources say
These are claims from the linked pages, not our measurements.
- Vendor-reported (DeepSeek, Hugging Face card): Terminal-Bench 4.0 42.3 and DeepSWE v1.1 77.0 for 1017, against 31.2 and 74.2 for the Sep 10 V4.1-Flash. These are vendor-run figures we did not verify.
Notes
- This is the same product line as DeepSeek-V4.1-Flash (Sep 10, 2026), which DeepSeek has replaced behind the deepseek-flash API name; we keep the earlier page for history.
What we tested
Nothing yet. We have not run DeepSeek-V4.1-Flash-1017 ourselves, so every fact on this page is Reported from the sources below. When we do test it, the scope will be stated here and in our test logs. See methodology.
Not verified, left out
We could not confirm these on a vendor page, so we do not state them.
- DeepSeek says the 1017 update did not change API pricing; we reuse the price list rather than a separate 1017 price page. The official list is in CNY and we do not convert currencies.
Sources
More from DeepSeek
- DeepSeek-V4.1-Flash (Sep 10, 2026)