Inference
LiveBench

LiveBench Language

Mean score over Connections word puzzles, unscrambling film plot summaries and repairing typos. English only.

ModelsModels with a result.55
Top resultBest result on this test.90.7%Claude Fable 5
Top-3 spreadPoints from first to third.1.3 pts
UpdatedDate the source changed the results.25 Jun 2026

Result and cost

020406080100$0.001$0.003$0.01$0.03$0.1$0.3$1RESULTUS DOLLARS PER TASK · LOG SCALE

Results

55 results
#ModelResultThe score from the source.CostUS dollars to run one task.
01Claude Fable 5max effort90.7%$0.446
02Claude Fable 5.1max effort89.5%$0.218
03GPT-6 Astra Max89.4%$0.222
04Claude Opus 5max effort88.7%$0.096
05Gemini 3.8 Flashhigh effort87.8%$0.088
06GPT-5.6 Sol Max87.7%$0.179
07GPT-5.5xhigh effort87.4%$0.19
08Claude Opus 5.5max effort86.3%$0.284
09Union Alpha85.9%$0
10Gemini 3.7 Flashhigh effort85.5%$0.031
11Gemini 3.5 Flashhigh effort84.6%$0.059
12Smaug Agentic84.4%$0.139
13Gemini 3.6 Flashhigh effort83.9%$0.051
14Grok 4.683.7%$0.035
15Claude Opus 4.6 Thinking Autohigh effort83.3%$0.101
16GPT-5.6 Terra Max82.9%$0.173
17Grok 4.582.8%$0.036
18Muse Spark 1.3xhigh effort82.8%$0.112
19GPT-5.4xhigh effort82.6%$0.19
20Claude Opus 4.5 Thinking 64Khigh effort81.3%$0.186
21DeepSeek V4.1 Flash Max81.2%$0.017
22DeepSeek V4 Flash Vision Exp80.4%$0.03
23Grok 4.7xhigh effort80.1%$0.305
24DeepSeek V4 Pro 042380.1%$0.014
25GLM-5.379.9%$0.241
26GPT-5.2high effort79.8%$0.042
27Qwen3.7 Max79.7%$0.037
28Qwen3.8 Max79.7%$0.135
29Claude Opus 4.8max effort79.7%$0.486
30Smaug Flash79.2%<$0.01
31Muse Spark 1.2xhigh effort78.6%$0.056
32Claude Opus 4.7xhigh effort77.9%$0.097
33Kimi K2.7 Code77.9%$0.035
34GLM-5.3-Flash77.3%<$0.01
35Smaug Mini77.0%$0.018
36MiniMax M376.8%$0.011
37GLM-5.276.2%$0.052
38Claude Sonnet 4.6 Thinking Automedium effort76.1%$0.061
39Kimi K2.6 Thinking75.1%$0.068
40Qwen3.6 Plus75.0%$0.032
41Claude Sonnet 5xhigh effort75.0%$0.127
42DeepSeek V4 Flash 042374.6%<$0.01
43Qwen3.8-Flash-Next74.6%$0.025
44Qwen3.8-27B74.3%$0.032
45Muse Spark 1.1xhigh effort74.3%$0.045
46GPT-5.2-Codex73.7%$0.071
47Grok 4.373.6%$0.023
48Inklingxhigh effort73.5%$0.085
49GPT-5.6 Luna Max72.6%$0.097
50Grok Build 0.172.5%<$0.01
51Gemini 3.5 Flash-Litehigh effort71.8%$0.016
52GPT-5.4 minixhigh effort71.0%$0.13
53Ox Alpha Max66.1%$0
54Qwen3.6-27B63.3%$0.048
55GPT-5.4 nanoxhigh effort62.5%$0.037