DeepSeek V4.1 Flash 1017 on BenchGecko · open weights under MIT, 1M context, 15 provider endpoints
DeepSeek released V4.1 Flash 1017 on 2026-10-17 with MIT weights. BenchGecko lists it at $0.30 input and $1.20 output per M tokens and tracks 15 provider endpoints.
What launched
DeepSeek released DeepSeek V4.1 Flash 1017 on 2026-10-17. It is an updated snapshot of DeepSeek V4.1 Flash, which shipped on 2026-09-10. DeepSeek serves it under the API name deepseek-flash, and the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are now answered by the same snapshot at the Flash price.
The weights are public on Hugging Face as deepseek-ai/DeepSeek-V4.1-Flash-1017 under the MIT licence.
Specs
From DeepSeek's API documentation and the BenchGecko model row:
- Context · 1M tokens, stored as 1,048,576 on BenchGecko
- Max output · 384K tokens
- Inputs · text and image
- Output · text
- Modes · thinking (default) and non-thinking
- Weights · open, MIT licence
Pricing on BenchGecko
BenchGecko lists the model at $0.30 per M input tokens and $1.20 per M output tokens, with cache reads at $0.006. These match DeepSeek's own peak rates. DeepSeek bills half that outside peak hours, so $0.15 input and $0.60 output. Peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays.
The previous V4.1 Flash row lists at $0.0472 input and $0.12 output per M tokens.
Provider spread
BenchGecko tracks 15 OpenRouter endpoints for the new snapshot. Six of them, DeepSeek included, charge the $0.30 and $1.20 list price. The cheapest input rate is Morph at $0.058 per M tokens in fp8. DeepInfra flex is next at $0.0784 input and $0.2352 output. InferenceNet caps context at 262,144 tokens. Every other endpoint serves the full 1,048,576.
Full table · provider prices.
Gecko Tests
The model has taken 3 Gecko Tests so far, for a Gecko score of 35 and rank 28.
- Who Are You · grade E · sometimes says it is OpenAI (finding)
- Knowledge Horizon · grade B · knows news up to May 2026
- Tokenizer Tax · grade D · 68% more tokens outside English
No public benchmark scores are attached to this snapshot yet. The earlier V4.1 Flash row carries 23 benchmark scores.
Sources
Source: BenchGecko
Written automatically from BenchGecko's tracked data for this period; every number comes from the changelog, model, mindshare and pricing datasets. Checked against the source data before publishing.