Inference
Artificial Analysis / Mercor

APEX-Agents-AA

Artificial Analysis' implementation of the APEX-Agents benchmark for long-horizon professional-services agent tasks.

ModelsModels with a result.26
Top resultBest result on this test.47.1%Gemini 3.5 Flash
Top-3 spreadPoints from first to third.8.2 pts
YearYear of release.2026

Result and price

020406080100$0.1$0.3$1$3$10$30$100RESULTOUTPUT PRICE PER 1M TOKENS · LOG SCALE

Results

26 results
#ModelResultThe score from the source.
01Gemini 3.5 Flash47.1%
02Kimi K341.3%
03GPT-5.6 Terra38.9%
04GPT-5.537.7%
05GPT-5.6 Luna35.8%
06GLM-5.233.7%
07GPT-5.433.3%
08Claude Opus 4.6 (Adaptive)33.0%
09Gemini 3.1 Pro32.0%
10Apodex 1.131.2%
10Apodex 1.1 Mini31.2%
12Kimi K2.628.5%
13GPT-5.4 mini28.2%
14GPT-5.4 nano24.9%
15DeepSeek V4 Pro 081324.3%
16Qwen3.7 Plus22.4%
17Grok 4.317.0%
18Step 3.7 Flash14.8%
19GLM-514.5%
20Kimi K2.5 (Reasoning)11.5%
20Kimi K2.511.5%
22MiniMax M2.710.6%
23GPT-OSS 120B3.1%
24MiMo-V2.5-Pro2.4%
25Nemotron 3 Super 100B1.8%
26GPT-OSS 20B0.7%