Kimi K2.5 or Qwen3.6 27B: which to choose?

What are we comparing?

This comparison pits two open-weight models that you can install on your own servers against each other: Kimi K2.5 and Qwen3.6 27B, one from Moonshot AI, the other from Alibaba Cloud (Qwen). The first dates from January 27, 2026, the second from April 22, 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, Kimi K2.5 scores 78.1 out of 100 and Qwen3.6 27B 72.2. Kimi K2.5 leads by 5.9 points.

On the 2 benchmarks reported for both, Kimi K2.5 wins 2 times and Qwen3.6 27B 0:

• GPQA Diamond (PhD-level science questions): Kimi K2.5 87.6%, Qwen3.6 27B 85.9%
• OTIS Mock AIME (competition mathematics problems): Kimi K2.5 92.2%, Qwen3.6 27B 91.1%

Budget

Kimi K2.5 costs $0.58 per million input tokens and $3 per million output tokens; Qwen3.6 27B costs $0.33 and $2.7.

For a typical agent task (read 20,000 tokens, write 2,000), expect $0.018 with Kimi K2.5 versus $0.012 with Qwen3.6 27B. Qwen3.6 27B is 1.5 times cheaper, a gap that adds up quickly over 1,000 tasks: $5.7 difference.

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

Context and capabilities

Both models have the same context window: 262,144 tokens, roughly 393 pages.

Reasoning mode: Kimi K2.5 yes, Qwen3.6 27B yes. Tool calling: Kimi K2.5 yes, Qwen3.6 27B yes. Image input: Kimi K2.5 yes, Qwen3.6 27B yes.

Both publish their weights: you can host them yourself.

In short

Our reading of the data:

• Choose Kimi K2.5 for a higher capability index.
• Choose Qwen3.6 27B for a lower cost and a more recent release.

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