Skip to main content
benchlm.ai

Vals-hosted LiveCodeBench mirror (Vals LiveCodeBench mirror)

Vals AI implementation of LiveCodeBench with easy, medium, and hard task splits.

How BenchLM shows Vals LiveCodeBench mirror

BenchLM mirrors the public Vals AI Vals LiveCodeBench mirror leaderboard captured from https://www.vals.ai/benchmarks/lcb?suggested=open-weights-table and updated by Vals on June 17, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

Vals LiveCodeBench mirror is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.

122 Vals rows4 task viewspublic datasetTasks: Overall, Easy, Medium, HardDisplay only
Vals LiveCodeBench mirror on Vals AIVals methodologyVals home

Vals LiveCodeBench score on Vals LiveCodeBench mirror — June 17, 2026

BenchLM mirrors the published vals livecodebench score view for Vals LiveCodeBench mirror. Claude Fable 5 leads the public snapshot at 89.78% , followed by Gemini 3.1 Pro Preview (88.48%) and GPT-5.2 Codex (87.99%). BenchLM does not use these results to rank models overall.

1

Claude Fable 5

Anthropic

anthropic/claude-fable-5

89.78%
Overall —
2

Gemini 3.1 Pro Preview

Google

google/gemini-3.1-pro-preview

88.48%
Overall —
3

GPT-5.2 Codex

OpenAI

openai/gpt-5.2-codex

87.99%
Overall —
122 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated June 17, 2026

The published Vals LiveCodeBench mirror snapshot is tightly clustered at the top: Claude Fable 5 sits at 89.78%, while the third row is only 1.79 points behind. The broader top-10 spread is 3.17 points, so many of the published scores sit in a relatively narrow band.

122 models have been evaluated on Vals LiveCodeBench mirror. The benchmark falls in the External benchmark mirrors category. BenchLM tracks this category separately from its weighted global scoring system, so these results are best compared on the dedicated Korean benchmark views. Vals LiveCodeBench mirror is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About Vals LiveCodeBench mirror

Year

2026

Tasks

Coding problem difficulty splits

Format

Accuracy score

Difficulty

Contamination-resistant coding problems

BenchLM keeps this separate from its canonical LiveCodeBench rows because it is a Vals-hosted implementation snapshot.

Vals LiveCodeBenchPublic benchmark source

BenchLM freshness & provenance

Version

Vals LiveCodeBench mirror 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Vals LiveCodeBench score table (122 models)

1
Claude Fable 5anthropic/claude-fable-5
89.78%
2
Gemini 3.1 Pro Previewgoogle/gemini-3.1-pro-preview
88.48%
3
GPT-5.2 Codexopenai/gpt-5.2-codex
87.99%
4
Claude Opus 4.8anthropic/claude-opus-4-8
87.82%
5
Gemini 3.5 Flashgoogle/gemini-3.5-flash
87.60%
6
DeepSeek V4 Prodeepseek/deepseek-v4-pro
87.48%
7
GPT-5.3 Codexopenai/gpt-5.3-codex
87.31%
8
Qwen3.7 Maxalibaba/qwen3.7-max
87.06%
9
Kimi K2.6kimi/kimi-k2.6
86.77%
10
GPT-5 Miniopenai/gpt-5-mini-2025-08-07
86.61%
11
GPT-5.1openai/gpt-5.1-2025-11-13
86.49%
12
Gemini 3 Pro Previewgoogle/gemini-3-pro-preview
86.41%
13
Nemotron 3 Ultra 550b A55bnvidia/nemotron-3-ultra-550b-a55b
85.98%
14
Qwen3.6 Plusalibaba/qwen3.6-plus
85.95%
15
GPT-5openai/gpt-5-2025-08-07
85.91%
16
Gemini 3 Flash Previewgoogle/gemini-3-flash-preview
85.59%
17
GPT-5.1 Codexopenai/gpt-5.1-codex
85.55%
18
GPT-5.2openai/gpt-5.2-2025-12-11
85.36%
19
Qwen3.5 Plus Thinkingalibaba/qwen3.5-plus-thinking
85.33%
20
GPT-5.5openai/gpt-5.5
85.30%
21
Claude Opus 4.7anthropic/claude-opus-4-7
85.07%
22
GPT-5 Codexopenai/gpt-5-codex
84.72%
23
Claude Opus 4.6 Thinkinganthropic/claude-opus-4-6-thinking
84.68%
24
Grok 4.3grok/grok-4.3
84.49%
25
Grok 4.20 0309 Reasoninggrok/grok-4.20-0309-reasoning
84.27%
26
GPT-5.4openai/gpt-5.4-2026-03-05
84.14%
27
GPT-5.4 Nanoopenai/gpt-5.4-nano-2026-03-17
84.01%
28
O3openai/o3-2025-04-16
83.91%
29
Kimi K2.5 Thinkingkimi/kimi-k2.5-thinking
83.87%
30
Claude Opus 4.5 20251101 Thinkinganthropic/claude-opus-4-5-20251101-thinking
83.67%
31
GPT-5.1 Codex Maxopenai/gpt-5.1-codex-max
83.56%
32
Qwen3.5 Flashalibaba/qwen3.5-flash
83.28%
33
Grok 4 0709grok/grok-4-0709
83.25%
34
GPT Oss 120bfireworks/gpt-oss-120b
83.23%
35
GLM 4.7zai/glm-4.7
82.23%
36
O4 Miniopenai/o4-mini-2025-04-16
82.21%
37
MiniMax M3minimax/MiniMax-M3
82.15%
38
Claude Sonnet 4.6anthropic/claude-sonnet-4-6
82.09%
39
Kimi K2.7 Codekimi/kimi-k2.7-code
82.05%
40
GLM 5 Thinkingzai/glm-5-thinking
81.87%
41
MiniMax M2.1minimax/MiniMax-M2.1
81.76%
42
Mimo V2.5xiaomi/mimo-v2.5
81.51%
43
GPT-5.4 Miniopenai/gpt-5.4-mini-2026-03-17
81.47%
44
GLM 5.1zai/glm-5.1
81.38%
45
Mimo V2.5 Proxiaomi/mimo-v2.5-pro
81.35%
46
GLM 4.6zai/glm-4.6
81.04%
47
DeepSeek V3p2 Thinkingfireworks/deepseek-v3p2-thinking
80.69%
48
Grok 4.1 Fast Reasoninggrok/grok-4-1-fast-reasoning
80.64%
49
GPT Oss 20bfireworks/gpt-oss-20b
80.39%
50
Gemini 3.1 Flash Lite Previewgoogle/gemini-3.1-flash-lite-preview
80.12%
51
MiniMax M2.7minimax/MiniMax-M2.7
79.93%
52
Command A Plus 05 2026cohere/command-a-plus-05-2026
79.38%
53
MiniMax M2.5minimax/MiniMax-M2.5
79.21%
54
Gemini 2.5 Pro Preview 03 25google/gemini-2.5-pro-preview-03-25
79.16%
55
Grok 4 Fast Reasoninggrok/grok-4-fast-reasoning
78.97%
56
Qwen3 Maxalibaba/qwen3-max
78.22%
57
Grok 3 Mini Fast High Reasoninggrok/grok-3-mini-fast-high-reasoning
76.22%
58
Gemini 2.5 Flash Preview 09 2025 Thinkinggoogle/gemini-2.5-flash-preview-09-2025-thinking
76.21%
59
Gemini 2.5 Flash Preview 09 2025google/gemini-2.5-flash-preview-09-2025
75.06%
60
Claude Opus 4.5anthropic/claude-opus-4-5-20251101
75.03%
61
Magistral Medium 2509mistralai/magistral-medium-2509
74.86%
62
Claude Sonnet 4.5 20250929 Thinkinganthropic/claude-sonnet-4-5-20250929-thinking
73.00%
63
Magistral Small 2509mistralai/magistral-small-2509
72.13%
64
O3 Miniopenai/o3-mini-2025-01-31
71.48%
65
Gemini 2.5 Flash Lite Preview 09 2025 Thinkinggoogle/gemini-2.5-flash-lite-preview-09-2025-thinking
71.39%
66
Qwen3 235b A22bfireworks/qwen3-235b-a22b
70.62%
67
Moonshotai Kimi K2 Instructtogether/moonshotai/Kimi-K2-Instruct
70.45%
68
DeepSeek R1fireworks/deepseek-r1
70.22%
69
GPT-5 Nanoopenai/gpt-5-nano-2025-08-07
70.22%
70
Claude Opus 4 20250514 Thinkinganthropic/claude-opus-4-20250514-thinking
70.19%
71
DeepSeek V3p2fireworks/deepseek-v3p2
69.86%
72
GLM 5.2zai/glm-5.2
69.50%
73
Gemini 2.5 Flash Lite Preview 09 2025google/gemini-2.5-flash-lite-preview-09-2025
67.67%
74
GLM 4.5zai/glm-4.5
67.45%
75
Qwen3 Max Previewalibaba/qwen3-max-preview
66.91%
76
Laguna Xs.2poolside/laguna-xs.2
66.61%
77
Claude Opus 4.1 20250805 Thinkinganthropic/claude-opus-4-1-20250805-thinking
66.46%
78
Grok 3 Mini Fast Low Reasoninggrok/grok-3-mini-fast-low-reasoning
66.27%
79
DeepSeek V3 0324fireworks/deepseek-v3-0324
65.48%
80
Claude Opus 4.1anthropic/claude-opus-4-1-20250805
64.56%
81
Kimi K2 Thinkingkimi/kimi-k2-thinking
63.15%
82
Claude Opus 4anthropic/claude-opus-4-20250514
62.63%
83
Claude Sonnet 4 20250514 Thinkinganthropic/claude-sonnet-4-20250514-thinking
62.39%
84
Grok Code Fast 1grok/grok-code-fast-1
61.97%
85
Laguna M.1poolside/laguna-m.1
61.44%
86
Claude 3.7 Sonnet 20250219 Thinkinganthropic/claude-3-7-sonnet-20250219-thinking
60.44%
87
Claude Sonnet 4anthropic/claude-sonnet-4-20250514
59.67%
88
Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561 Thinkingtogether/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561-thinking
58.37%
89
GPT-4.1 Miniopenai/gpt-4.1-mini-2025-04-14
58.16%
90
Gemini 2.5 Flash Preview 04 17google/gemini-2.5-flash-preview-04-17
56.94%
91
Claude 3.7 Sonnetanthropic/claude-3-7-sonnet-20250219
56.66%
92
Mistral Large 2512mistralai/mistral-large-2512
55.34%
93
GPT-4.1openai/gpt-4.1-2025-04-14
54.67%
94
Grok 3grok/grok-3
52.90%
95
Devstral 2512mistralai/devstral-2512
51.84%
96
O1openai/o1-2024-12-17
50.26%
97
Claude 3.5 Sonnetanthropic/claude-3-5-sonnet-20241022
49.63%
98
Llama4 Maverick Instruct Basicfireworks/llama4-maverick-instruct-basic
47.25%
99
Gemini 2.5 Flash Preview 04 17 Thinkinggoogle/gemini-2.5-flash-preview-04-17-thinking
46.87%
100
Grok 4 Fast Non Reasoninggrok/grok-4-fast-non-reasoning
46.09%
101
Mistral Medium 2505mistralai/mistral-medium-2505
44.84%
102
Gemini 2.0 Flash 001google/gemini-2.0-flash-001
43.61%
103
GPT-4oopenai/gpt-4o-2024-11-20
43.44%
104
Labs Devstral Small 2512mistralai/labs-devstral-small-2512
43.18%
105
GPT-4.1 Nanoopenai/gpt-4.1-nano-2025-04-14
42.72%
106
Grok 4.1 Fast Non Reasoninggrok/grok-4-1-fast-non-reasoning
42.62%
107
Claude 3.5 Haikuanthropic/claude-3-5-haiku-20241022
41.92%
108
Gemini 1.5 Pro 002google/gemini-1.5-pro-002
41.72%
109
Claude Haiku 4.5 20251001 Thinkinganthropic/claude-haiku-4-5-20251001-thinking
41.17%
110
Grok 2 1212grok/grok-2-1212
38.68%
111
Meta Llama Llama 4 Scout 17B 16E Instructtogether/meta-llama/Llama-4-Scout-17B-16E-Instruct
38.54%
112
Mistral Large 2411mistralai/mistral-large-2411
37.09%
113
Gemini 1.5 Flash 002google/gemini-1.5-flash-002
36.91%
114
Meta Llama Llama 3.3 70B Instruct Turbotogether/meta-llama/Llama-3.3-70B-Instruct-Turbo
36.34%
115
Langston Nim Nvidia Llama 3.3 Nemotron Super 49b V1 42e84561together/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561
36.31%
116
Command A 03 2025cohere/command-a-03-2025
35.07%
117
Mistral Small 2503mistralai/mistral-small-2503
31.82%
118
GPT-4o Miniopenai/gpt-4o-mini-2024-07-18
26.42%
119
Jamba Large 1.6ai21labs/jamba-large-1.6
22.32%
120
Command R Pluscohere/command-r-plus
18.24%
121
Mistral Small 2402mistralai/mistral-small-2402
15.78%
122
Jamba Mini 1.6ai21labs/jamba-mini-1.6
9.92%

FAQ

What does Vals LiveCodeBench mirror measure?

Vals AI implementation of LiveCodeBench with easy, medium, and hard task splits.

Which model leads the published Vals LiveCodeBench mirror snapshot?

Claude Fable 5 currently leads the published Vals LiveCodeBench mirror snapshot with 89.78% vals livecodebench score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on Vals LiveCodeBench mirror?

122 AI models are included in BenchLM's mirrored Vals LiveCodeBench mirror snapshot, based on the public leaderboard captured on June 17, 2026.

Last updated: June 17, 2026 · mirrored from the public benchmark leaderboard

The AI models change fast. We track them for you.

For engineers, researchers, and the plain curious — a weekly brief on new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.