ModelsLiveBench

LiveBench

LiveBench is an objective, contamination-resistant benchmark: questions are refreshed so models cannot memorize the test set. Pick a category below: Overall plus Reasoning, Coding, Agentic Coding, Math, Data Analysis, Language, and Instruction Following (IF). Overall is the mean of category averages. This page mirrors the LiveBench-2026-06-25 public release.

Newer flagships (including Grok 4.7) may be absent until the next LiveBench refresh — a missing row is not a score of zero. For human preference across multiple Arena boards, see LMArena boards. For the Artificial Analysis Index cut used elsewhere on the site, see Models. Also: Price vs value · Pricing history.

Snapshot as of Jun 25, 2026 (LiveBench-2026-06-25). Not investment advice. Figures lag the labs. Prefer livebench.ai for the latest release.

Overall

Mean of category averages. Higher is better.

Bars scaled to this board's score range (not from zero), so small gaps show up.

Frontier closedOpen weight

Claude Fable 5.183.4
Claude Fable 583
GPT-6 Astra82.2
Muse Spark 1.381.6
DeepSeek V4.1 Flash81.1
GPT-5.6 Sol81
GPT-5.580.2
Claude Opus 580.1
Kimi K379.2
Gemini 3.7 Flash78.8
Qwen3.8 Max78.5
Grok 4.678
Muse Spark 1.278
GPT-5.6 Terra77.9
Gemini 3.1 Pro Preview77
GPT-5.4 Mini66.4
Qwen3.6 27B64
Grok 4.362.3

Full table

Sortable detail across all categories. Models missing a category score show an em dash in that column.

Sort

#1 Claude Fable 5.1
Anthropic · Max Effort
Overall
83.4
Cost / task
$1.212
Reasoning
91.7
Coding
86.4
Agentic Coding
66.1
Math
97
Data Analysis
80.3
Language
89.5
IF
73
#2 Claude Fable 5
Anthropic · Max Effort
Overall
83
Cost / task
$1.439
Reasoning
89.7
Coding
86
Agentic Coding
62.2
Math
96
Data Analysis
80.5
Language
90.7
IF
75.8
#3 GPT-6 Astra
OpenAI · Max Effort
Overall
82.2
Cost / task
$0.736
Reasoning
92.7
Coding
80.4
Agentic Coding
57.3
Math
96.8
Data Analysis
83
Language
89.4
IF
75.6
#4 Muse Spark 1.3
Meta · xHigh
Overall
81.6
Cost / task
$0.219
Reasoning
89.7
Coding
81.1
Agentic Coding
64.1
Math
95.9
Data Analysis
79.6
Language
82.8
IF
78
#5 DeepSeek V4.1 Flash
DeepSeek · Max · Open
Overall
81.1
Cost / task
$0.029
Reasoning
86.7
Coding
80
Agentic Coding
77.3
Math
93.3
Data Analysis
79.3
Language
81.2
IF
70
#6 GPT-5.6 Sol
OpenAI · Max Effort
Overall
81
Cost / task
$0.515
Reasoning
91.7
Coding
83.9
Agentic Coding
56.2
Math
96.2
Data Analysis
79.8
Language
87.7
IF
71.8
#7 GPT-5.5
OpenAI · xHigh Thinking
Overall
80.2
Cost / task
$0.435
Reasoning
89.7
Coding
82.1
Agentic Coding
54
Math
95.9
Data Analysis
81.6
Language
87.4
IF
70.7
#8 Claude Opus 5
Anthropic · Max Effort Thinking
Overall
80.1
Cost / task
$0.699
Reasoning
91.2
Coding
81.4
Agentic Coding
65.2
Math
95.7
Data Analysis
74.6
Language
88.7
IF
63.8
#9 Kimi K3
Moonshot · Open
Overall
79.2
Cost / task
$0.348
Reasoning
90.7
Coding
81.4
Agentic Coding
62.2
Math
84.4
Data Analysis
78.7
Language
85.5
IF
71.4
#10 Gemini 3.7 Flash
Google · High
Overall
78.8
Cost / task
$0.157
Reasoning
87.8
Coding
78.9
Agentic Coding
58.3
Math
93.5
Data Analysis
68
Language
85.5
IF
79.9
#11 Qwen3.8 Max
Alibaba · Open
Overall
78.5
Cost / task
$0.275
Reasoning
88.2
Coding
72.9
Agentic Coding
64.6
Math
91.3
Data Analysis
78.4
Language
79.7
IF
74.1
#12 Grok 4.6
xAI
Overall
78
Cost / task
$0.207
Reasoning
90.5
Coding
76.8
Agentic Coding
57
Math
92.6
Data Analysis
73.9
Language
83.7
IF
71.9
#14 Muse Spark 1.2
Meta · xHigh
Overall
78
Cost / task
—
Reasoning
—
Coding
—
Agentic Coding
—
Math
—
Data Analysis
—
Language
—
IF
—
#13 GPT-5.6 Terra
OpenAI · Max Effort
Overall
77.9
Cost / task
$0.352
Reasoning
90.6
Coding
78.2
Agentic Coding
54.9
Math
94.9
Data Analysis
79.3
Language
82.9
IF
64.6
#15 Gemini 3.1 Pro Preview
Google · High
Overall
77
Cost / task
$0.286
Reasoning
84
Coding
76.5
Agentic Coding
44.1
Math
91
Data Analysis
78.5
Language
85.4
IF
79.1
#16 GPT-5.4 Mini
OpenAI · xHigh
Overall
66.4
Cost / task
$0.334
Reasoning
71.3
Coding
71.6
Agentic Coding
41.7
Math
78.5
Data Analysis
70.8
Language
71
IF
59.8
#17 Qwen3.6 27B
Alibaba · Open
Overall
64
Cost / task
$0.202
Reasoning
70.3
Coding
71.8
Agentic Coding
39.3
Math
79.9
Data Analysis
70.4
Language
63.3
IF
53.2
#18 Grok 4.3
xAI
Overall
62.3
Cost / task
—
Reasoning
—
Coding
—
Agentic Coding
—
Math
—
Data Analysis
—
Language
—
IF
—


1	Claude Fable 5.1 Anthropic · Max Effort	83.4	91.7	86.4	66.1	97	80.3	89.5	73	$1.212
2	Claude Fable 5 Anthropic · Max Effort	83	89.7	86	62.2	96	80.5	90.7	75.8	$1.439
3	GPT-6 Astra OpenAI · Max Effort	82.2	92.7	80.4	57.3	96.8	83	89.4	75.6	$0.736
4	Muse Spark 1.3 Meta · xHigh	81.6	89.7	81.1	64.1	95.9	79.6	82.8	78	$0.219
5	DeepSeek V4.1 Flash DeepSeek · Max	81.1	86.7	80	77.3	93.3	79.3	81.2	70	$0.029
6	GPT-5.6 Sol OpenAI · Max Effort	81	91.7	83.9	56.2	96.2	79.8	87.7	71.8	$0.515
7	GPT-5.5 OpenAI · xHigh Thinking	80.2	89.7	82.1	54	95.9	81.6	87.4	70.7	$0.435
8	Claude Opus 5 Anthropic · Max Effort Thinking	80.1	91.2	81.4	65.2	95.7	74.6	88.7	63.8	$0.699
9	Kimi K3 Moonshot	79.2	90.7	81.4	62.2	84.4	78.7	85.5	71.4	$0.348
10	Gemini 3.7 Flash Google · High	78.8	87.8	78.9	58.3	93.5	68	85.5	79.9	$0.157
11	Qwen3.8 Max Alibaba	78.5	88.2	72.9	64.6	91.3	78.4	79.7	74.1	$0.275
12	Grok 4.6 xAI	78	90.5	76.8	57	92.6	73.9	83.7	71.9	$0.207
14	Muse Spark 1.2 Meta · xHigh	78	—	—	—	—	—	—	—	—
13	GPT-5.6 Terra OpenAI · Max Effort	77.9	90.6	78.2	54.9	94.9	79.3	82.9	64.6	$0.352
15	Gemini 3.1 Pro Preview Google · High	77	84	76.5	44.1	91	78.5	85.4	79.1	$0.286
16	GPT-5.4 Mini OpenAI · xHigh	66.4	71.3	71.6	41.7	78.5	70.8	71	59.8	$0.334
17	Qwen3.6 27B Alibaba	64	70.3	71.8	39.3	79.9	70.4	63.3	53.2	$0.202
18	Grok 4.3 xAI	62.3	—	—	—	—	—	—	—	—

LiveBench

Full table

Know what matters in AI.