livebench reasoning spatial 2026_06_25 Benchmark: Scores and Sources

Published result; benchmark version and evaluation conditions remain in the id and result note.

pillar: reasoning · weight 1 within pillar · unit: % (higher is better) · official board ↗

Row	Model	Result	Normalized (0–100)	Confidence
1	Qwen3.6 27B Alibaba / Qwen	100%	100.0	53% confidence 53 percent, Medium
2	Qwen3.8 Flash Next Alibaba / Qwen	100%	100.0	21% confidence 21 percent, Low
3	Qwen3.8 Max Alibaba / Qwen	100%	100.0	85% confidence 85 percent, High
4	Claude Fable 5.1 Anthropic	100%	100.0	100% confidence 100 percent, Full
5	Claude Opus 4.7 Anthropic	100%	100.0	100% confidence 100 percent, Full
6	Claude Opus 5 Anthropic	100%	100.0	100% confidence 100 percent, Full
7	Claude Sonnet 4.6 Anthropic	100%	100.0	100% confidence 100 percent, Full
8	Claude Sonnet 5 Anthropic	100%	100.0	93% confidence 93 percent, High
9	Gemini 3.6 Flash Google	100%	100.0	100% confidence 100 percent, Full
10	Gemini 3.7 Flash Google	100%	100.0	100% confidence 100 percent, Full
11	Gemini 3.8 Flash Google	100%	100.0	100% confidence 100 percent, Full
12	Muse Spark 1.1 Meta	100%	100.0	61% confidence 61 percent, Medium
13	Muse Spark 1.3 Meta	100%	100.0	93% confidence 93 percent, High
14	Kimi K3 Moonshot AI	100%	100.0	100% confidence 100 percent, Full
15	GPT-5.6 Sol OpenAI	100%	100.0	100% confidence 100 percent, Full
16	GPT-5.6 Terra OpenAI	100%	100.0	100% confidence 100 percent, Full
17	Grok 4.5 xAI	100%	100.0	100% confidence 100 percent, Full
18	Qwen3.6 Plus Alibaba / Qwen	98%	98.0	80% confidence 80 percent, High
19	Claude Opus 4.6 Anthropic	98%	98.0	100% confidence 100 percent, Full
20	Claude Opus 4.8 Anthropic	98%	98.0	100% confidence 100 percent, Full
21	Claude Opus 5.5 Anthropic	98%	98.0	100% confidence 100 percent, Full
22	Claude Sonnet 5.5 Anthropic	98%	98.0	93% confidence 93 percent, High
23	DeepSeek V4.1 Flash DeepSeek	98%	98.0	69% confidence 69 percent, Medium
24	Gemini 3.1 Pro Preview Google	98%	98.0	100% confidence 100 percent, Full
25	Nemotron 3 Ultra 550B A55B nvidia	98%	98.0	100% confidence 100 percent, Full
26	GPT-5.4 OpenAI	98%	98.0	100% confidence 100 percent, Full
27	GPT-5.5 OpenAI	98%	98.0	100% confidence 100 percent, Full
28	GPT-6 Astra OpenAI	98%	98.0	100% confidence 100 percent, Full
29	GPT-6.1 Sol OpenAI	98%	98.0	100% confidence 100 percent, Full
30	Grok 4.3 xAI	98%	98.0	80% confidence 80 percent, High
31	Grok 4.6 xAI	98%	98.0	100% confidence 100 percent, Full
32	Grok Build 0.1 xAI	98%	98.0	16% confidence 16 percent, Low
33	Qwen3.7 Max Alibaba / Qwen	96%	96.0	53% confidence 53 percent, Medium
34	Qwen3.8 27B Alibaba / Qwen	96%	96.0	69% confidence 69 percent, Medium
35	Claude Fable 5 Anthropic	96%	96.0	100% confidence 100 percent, Full
36	Claude Opus 4.5 Anthropic	96%	96.0	100% confidence 100 percent, Full
37	DeepSeek V4 Flash DeepSeek	96%	96.0	53% confidence 53 percent, Medium
38	DeepSeek V4 Flash Vision Exp DeepSeek	96%	96.0	21% confidence 21 percent, Low
39	DeepSeek V4 Pro 0813 DeepSeek	96%	96.0	69% confidence 69 percent, Medium
40	Gemini 3.5 Flash Google	96%	96.0	100% confidence 100 percent, Full
41	Muse Spark 1.2 Meta	96%	96.0	53% confidence 53 percent, Medium
42	GPT-5.2 OpenAI	96%	96.0	100% confidence 100 percent, Full
43	GPT-5.6 Luna OpenAI	96%	96.0	100% confidence 100 percent, Full
44	GPT-6 Luna OpenAI	96%	96.0	100% confidence 100 percent, Full
45	GPT-6 Sol OpenAI	96%	96.0	100% confidence 100 percent, Full
46	GLM-5.3 Z.ai	96%	96.0	93% confidence 93 percent, High
47	GLM-5.3-Flash Z.ai	96%	96.0	100% confidence 100 percent, Full
48	DeepSeek V4 Pro DeepSeek	94%	94.0	85% confidence 85 percent, High
49	MiniMax-M3 minimax	94%	94.0	100% confidence 100 percent, Full
50	Mistral Large 4 mistral	94%	94.0	32% confidence 32 percent, Low
51	Kimi K2.6 Moonshot AI	94%	94.0	85% confidence 85 percent, High
52	GPT-5.2 Codex OpenAI	94%	94.0	43% confidence 43 percent, Low
53	Inkling thinkingmachines	94%	94.0	100% confidence 100 percent, Full
54	Grok 4.7 xAI	94%	94.0	100% confidence 100 percent, Full
55	GLM-5.2 Z.ai	94%	94.0	100% confidence 100 percent, Full
56	Claude Haiku 5.5 Anthropic	92%	92.0	100% confidence 100 percent, Full
57	Kimi K2.7 Code Moonshot AI	92%	92.0	48% confidence 48 percent, Low
58	DeepSeek V4 Flash 0731 DeepSeek	90%	90.0	69% confidence 69 percent, Medium
59	Gemini 3.5 Flash Lite Google	90%	90.0	100% confidence 100 percent, Full
60	GPT-5.4 mini OpenAI	88%	88.0	100% confidence 100 percent, Full
61	GPT-5.4 nano OpenAI	82%	82.0	100% confidence 100 percent, Full

Results are as published by the source behind each value (hover or tap the number). Benchmark scores are shown individually for every source/variant (a model may have multiple rows), and vary by version, harness and date; the normalized column uses the method’s fixed 0–100 scales and feeds the reasoning pillar of the SI Score.