gpt-oss-120b vs kimi-k2-0905-preview

Capability for the money

gpt-oss-120b is 2.4× cheaper than kimi-k2-0905-preview and scores higher on following instructions (IFBench) and coding (LiveCodeBench) — there is no case for paying more here.

1. Following instructions

   extraction, classification, sticking to a format

   IFBench
   gpt-oss-120b69.0%

   kimi-k2-0905-preview41.7%

   gpt-oss-120b +27.3

2. Coding

   writing and fixing code

   LiveCodeBench
   gpt-oss-120b87.8%

   kimi-k2-0905-preview61.0%

   gpt-oss-120b +26.8

3. Tool use

   multi-step agent work against real tools

   τ²-bench
   gpt-oss-120b65.8%

   kimi-k2-0905-preview73.4%

   kimi-k2-0905-preview +7.6

4. Hard reasoning

   graduate-level science and multi-step analysis

   GPQA Diamond
   gpt-oss-120b78.2%

   kimi-k2-0905-preview76.7%

   gpt-oss-120b +1.5

Averaged over the benchmarks both models sat, gpt-oss-120b returns 167.1 points of pass rate per dollar of blended price against 58.8 — the ratio that decides whether the dearer model earns its rate.

Each figure is a pass rate: the share of that test set the model answered correctly, measured by a third party rather than self-reported by the lab. Rows are different tests of different difficulty, so compare a model against the other model, not one row against the next. And a benchmark is not your workload — it narrows the shortlist, it does not pick for you.