gpt-oss-120b vs Kimi K2 Thinking

Benchmarks, API pricing and specs, head to head. Data updated 2026-06-10.

gpt-oss-120b

OpenAI · Aug 2025

OpenAI's open-weights MoE reasoning model under Apache 2.0 — near o4-mini quality, runnable on a single 80GB GPU.

Kimi K2 Thinking

Moonshot AI · Nov 2025

67.6

A trillion-parameter open reasoning agent that can chain 200–300 tool calls — the open-weights agentic standout of late 2025.

The verdict

Kimi K2 Thinking wins 3 of the 3 benchmarks these models share, against 0 for gpt-oss-120b. On price the gap is dramatic: gpt-oss-120b works out roughly 5.4x cheaper per blended million tokens. Kimi K2 Thinking also takes 262K of context versus 131K for gpt-oss-120b.

Benchmark head-to-head 03

GPQA Diamond
80.1%84.5%
gpt-oss-120bKimi K2 Thinking
AIME 2025
~92.5%~94.5%
gpt-oss-120bKimi K2 Thinking
HLE
19%23.9%
gpt-oss-120bKimi K2 Thinking

Specs & pricing

gpt-oss-120bKimi K2 Thinking
modhub Index67.6
Input price / 1M$0.1$0.6
Output price / 1M$0.5$2.5
Context window131K262K
Max output131K
Open weightsyes (Apache 2.0)yes (Modified MIT)
Reasoning modelyesyes
Multimodal inputtexttext
Knowledge cutoffJun 2024Apr 2025
ReleasedAug 2025Nov 2025
Example monthly cost*$1.75$9.75

* 10M input + 1.5M output tokens per month at list prices, no caching. Green = better value on that row.

Frequently asked questions

Which is better, gpt-oss-120b or Kimi K2 Thinking?
Kimi K2 Thinking wins 3 of the 3 benchmarks these models share, against 0 for gpt-oss-120b. On price the gap is dramatic: gpt-oss-120b works out roughly 5.4x cheaper per blended million tokens. Kimi K2 Thinking also takes 262K of context versus 131K for gpt-oss-120b.
Which is cheaper, gpt-oss-120b or Kimi K2 Thinking?
gpt-oss-120b costs $0.1/$0.5 per million input/output tokens, while Kimi K2 Thinking costs $0.6/$2.5. For a typical workload of 10M input and 1.5M output tokens per month, that's $1.75 versus $9.75.
Which model is better for coding, gpt-oss-120b or Kimi K2 Thinking?
We don't yet track SWE-bench Verified results for both models; check their individual pages for coding-related scores.

More comparisons