Arena Document · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Opus 5	high	1516	–	–
2	Fable 5.1	max	1513	–	–
3	Opus 4.6	high	1507	–	–
4	Fable 5	max	1496	–	–
5	Opus 4.7	max	1495	–	–
6	GPT-5.5	xhigh	1486	–	–
7	GPT-5.6 Sol	xhigh	1483	–	–
8	Sonnet 4.6	max	1482	–	–
9	Opus 4.8	high	1475	–	–
10	GPT-5.6 Terra	xhigh	1472	–	–
11	GPT-5.4	xhigh	1471	–	–
12	Muse Spark 1.3	max	1471	–	–
13	GPT-6 Astra	max	1468	–	–
14	Sonnet 5	high	1466	–	–
15	Muse Spark 1.1	max	1465	–	–
16	Gemini 3.5 Flash	medium	1463	–	–
17	Opus 4.5	high	1462	–	–
18	GPT-5.6 Luna	xhigh	1457	–	–
19	Gemini 3.6 Flash	high	1456	–	–
20	Grok 4.6	high	1452	–	–
21	Grok 4.5	max	1452	–	–
22	Kimi K2.6	max	1451	–	–
23	Sonnet 4.5	high	1450	–	–
24	Qwen 3.7 Plus	high	1444	–	–
25	Gemini 3.1 Pro	high	1444	–	–
26	MiniMax M3	max	1435	–	–
27	Gemini 3 Pro	high	1434	–	–
28	Kimi K2.5	max	1430	–	–
29	Gemma 4 31B	max	1425	–	–
30	Gemini 2.5 Pro	high	1421	–	–
31	Haiku 4.5	xhigh	1420	–	–
32	Grok 4.20	xhigh	1416	–	–
33	GLM 5 V	high	1416	–	–
34	Gemini 3 Flash	high	1413	–	–
35	GPT-5.2	high	1405	–	–
36	GPT-5.1	high	1403	–	–
37	GPT-5.5 Instant	–	1403	–	–