DeepSeek V4.1 Flash 1017 vs Kimi K3

What are we comparing?

This comparison pits two open-weight models that you can install on your own servers against each other: DeepSeek V4.1 Flash 1017 and Kimi K3, one from DeepSeek, the other from Moonshot AI. The first dates from October 17, 2026, the second from July 16, 2026.

All figures below come from public sources checked on October 22, 2026: models.dev and OpenRouter for prices and capabilities, Epoch AI for scores.

What the benchmarks say

On our capability index, DeepSeek V4.1 Flash 1017 scores 90.6 out of 100 and Kimi K3 80.7. DeepSeek V4.1 Flash 1017 leads by 9.9 points.

On the 3 benchmarks reported for both, DeepSeek V4.1 Flash 1017 wins 2 times and Kimi K3 1:

• GPQA Diamond (PhD-level science questions): DeepSeek V4.1 Flash 1017 93.2%, Kimi K3 93.1%
• OTIS Mock AIME (competition mathematics problems): DeepSeek V4.1 Flash 1017 99.4%, Kimi K3 97.2%
• FrontierMath (very hard research-level mathematics): DeepSeek V4.1 Flash 1017 71.2%, Kimi K3 72.2%

Pricing

DeepSeek V4.1 Flash 1017 costs $0.15 per million input tokens and $0.60 per million output tokens; Kimi K3 costs $3 and $15.

For a typical agent task (read 20,000 tokens, write 2,000), expect $0.004 with DeepSeek V4.1 Flash 1017 versus $0.09 with Kimi K3. DeepSeek V4.1 Flash 1017 is 21.4 times cheaper, a gap that adds up quickly over 1,000 tasks: $85.8 difference.

Run the numbers with your own volumes in the agent cost calculator.

Features and limits

Both models have the same context window: 1,048,576 tokens, roughly 1,573 pages.

Reasoning mode: DeepSeek V4.1 Flash 1017 yes, Kimi K3 yes. Tool calling: DeepSeek V4.1 Flash 1017 yes, Kimi K3 yes. Image input: DeepSeek V4.1 Flash 1017 yes, Kimi K3 yes.

Both publish their weights: you can host them yourself.

In short

Our reading of the data:

• Choose DeepSeek V4.1 Flash 1017 for a higher capability index, a lower cost and a more recent release.

This verdict relies on public figures; it does not replace a trial on your own tasks.