Model Comparator

Compare up to four models using the same criteria.

DeepSeek V3.2 (Non-reasoning)
DeepSeek
K2 Horizon 3.7B
Institute of Foundation Models

Differences only

✓ Best published value among measured models. — Data unavailable. These metrics do not establish an overall ranking.

Model Comparator
Metrics	DeepSeek V3.2 (Non-reasoning) DeepSeek	K2 Horizon 3.7B Institute of Foundation Models
Intelligence	Best: 16	15.6
Coding	Best: 44.2	26.1
Math	59	—
Agentic	—	16.5
Input · $ / 1M tokens	$0.28	—
Output · $ / 1M tokens	$0.42	—
Context window	128K	Best: 524K
Reasoning	No	Yes
Open weights	Yes	Yes

Intelligence

DeepSeek V3.2 (Non-reasoning): Best: 16
K2 Horizon 3.7B: 15.6

Coding

DeepSeek V3.2 (Non-reasoning): Best: 44.2
K2 Horizon 3.7B: 26.1

Math

DeepSeek V3.2 (Non-reasoning): 59
K2 Horizon 3.7B: —

Agentic

DeepSeek V3.2 (Non-reasoning): —
K2 Horizon 3.7B: 16.5

Input · $ / 1M tokens

DeepSeek V3.2 (Non-reasoning): $0.28
K2 Horizon 3.7B: —

Output · $ / 1M tokens

DeepSeek V3.2 (Non-reasoning): $0.42
K2 Horizon 3.7B: —

Context window

DeepSeek V3.2 (Non-reasoning): 128K
K2 Horizon 3.7B: Best: 524K

Reasoning

DeepSeek V3.2 (Non-reasoning): No
K2 Horizon 3.7B: Yes

Open weights

DeepSeek V3.2 (Non-reasoning): Yes
K2 Horizon 3.7B: Yes

Each reasoning level remains a separate configuration.

Score and cost

Each line joins the reasoning levels of a compared model; the large point marks the configuration in the table. The top-right corner holds the best score for the lowest cost. Select another point to compare that level.

DeepSeek V3.2 (Non-reasoning)16 · $0.315

Not plotted for lack of data on these axes: K2 Horizon 3.7B.