Skip to main content

Vibe Code Bench v1.1 (Vibe Code Bench)

Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment.

How BenchLM shows Vibe Code Bench v1.1

BenchLM mirrors the public Vals AI Vibe Code Bench v1.1 leaderboard captured from https://www.vals.ai/benchmarks/vibe-code and updated by Vals on May 28, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

Vibe Code Bench v1.1 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.

47 Vals rows1 task viewsprivate datasetTasks: OverallDisplay only

Vibe Code score on Vibe Code Bench — May 28, 2026

BenchLM mirrors the published vibe code score view for Vibe Code Bench. Claude Opus 4.8 leads the public snapshot at 82.72% , followed by Claude Opus 4.7 (71.00%) and GPT-5.5 (69.85%). BenchLM does not use these results to rank models overall.

47 modelsCodingCurrentDisplay onlyUpdated May 28, 2026

The published Vibe Code Bench snapshot is tightly clustered at the top: Claude Opus 4.8 sits at 82.72%, while the third row is only 12.88 points behind. The broader top-10 spread is 32.79 points, so the benchmark still separates strong models even when the leaders cluster.

47 models have been evaluated on Vibe Code Bench. The benchmark falls in the Coding category. This category carries a 20% weight in BenchLM.ai's overall scoring system. Vibe Code Bench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About Vibe Code Bench

Year

2026

Tasks

End-to-end web application builds

Format

Full-stack app implementation benchmark

Difficulty

End-to-end software delivery

Vibe Code Bench v1.1 asks models to build full web apps with services such as Supabase, Stripe test mode, email, browsing, and file editing available. The score is overall application pass accuracy across private end-to-end app tasks.

BenchLM freshness & provenance

Version

Vibe Code Bench 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.

Vibe Code score table (47 models)

1
Claude Opus 4.8anthropic/claude-opus-4-8
82.72%
2
Claude Opus 4.7anthropic/claude-opus-4-7
71.00%
3
GPT-5.5openai/gpt-5.5
69.85%
4
GPT-5.4openai/gpt-5.4-2026-03-05
67.42%
5
GPT-5.3 Codexopenai/gpt-5.3-codex
61.77%
6
Claude Opus 4.6anthropic/claude-opus-4-6
57.57%
7
GPT-5.2openai/gpt-5.2-2025-12-11
53.50%
8
Claude Opus 4.6 Thinkinganthropic/claude-opus-4-6-thinking
53.50%
9
Claude Sonnet 4.6anthropic/claude-sonnet-4-6
51.48%
10
DeepSeek V4 Prodeepseek/deepseek-v4-pro
49.93%
11
Gemini 3.5 Flashgoogle/gemini-3.5-flash
48.68%
12
GPT-5.4 Miniopenai/gpt-5.4-mini-2026-03-17
47.97%
13
GPT-5.2 Codexopenai/gpt-5.2-codex
37.91%
14
Kimi K2.6 Thinkingkimi/kimi-k2.6-thinking
37.89%
15
Gemini 3.1 Pro Previewgoogle/gemini-3.1-pro-preview
32.03%
16
GLM 5.1 Thinkingzai/glm-5.1-thinking
31.46%
17
GPT-5.4 Nanoopenai/gpt-5.4-nano-2026-03-17
26.10%
18
Qwen3.6 Plusalibaba/qwen3.6-plus
25.57%
19
GPT-5.1openai/gpt-5.1-2025-11-13
24.61%
20
GLM 5 Thinkingzai/glm-5-thinking
23.36%
21
Claude Sonnet 4.5 20250929 Thinkinganthropic/claude-sonnet-4-5-20250929-thinking
22.62%
22
GPT-5.1 Codex Maxopenai/gpt-5.1-codex-max
22.17%
23
Claude Opus 4.5 20251101 Thinkinganthropic/claude-opus-4-5-20251101-thinking
20.63%
24
Gemini 3 Flash Previewgoogle/gemini-3-flash-preview
20.20%
25
GPT-5openai/gpt-5-2025-08-07
20.09%
26
Muse Sparkmeta/muse_spark
19.67%
27
Grok 4.3grok/grok-4.3
19.40%
28
Kimi K2.5 Thinkingkimi/kimi-k2.5-thinking
17.54%
29
Qwen3.5 Plus Thinkingalibaba/qwen3.5-plus-thinking
15.74%
30
MiniMax M2.5minimax/MiniMax-M2.5
14.85%
31
Gemini 3 Pro Previewgoogle/gemini-3-pro-preview
14.30%
32
GPT-5 Miniopenai/gpt-5-mini-2025-08-07
14.17%
33
GPT-5.1 Codexopenai/gpt-5.1-codex
13.12%
34
Qwen3.6 27balibaba/qwen3.6-27b
11.94%
35
MiniMax M2.7minimax/MiniMax-M2.7
11.93%
36
Qwen3.7 Maxalibaba/qwen3.7-max
11.42%
37
Claude Haiku 4.5 20251001 Thinkinganthropic/claude-haiku-4-5-20251001-thinking
11.39%
38
DeepSeek V3p2 Thinkingfireworks/deepseek-v3p2-thinking
5.11%
39
Grok 4.20 0309 Reasoninggrok/grok-4.20-0309-reasoning
4.06%
40
Qwen3 Maxalibaba/qwen3-max
3.51%
41
GLM 4.6zai/glm-4.6
3.09%
42
Grok 4.1 Fast Reasoninggrok/grok-4-1-fast-reasoning
1.20%
43
Gemini 2.5 Progoogle/gemini-2.5-pro
0.40%
44
Command A Plus 05 2026cohere/command-a-plus-05-2026
0.00%
45
Gemini 3.1 Flash Lite Previewgoogle/gemini-3.1-flash-lite-preview
0.00%
46
Grok 4 Fast Reasoninggrok/grok-4-fast-reasoning
0.00%
47
Mistral Small 2603mistralai/mistral-small-2603
0.00%

FAQ

What does Vibe Code Bench measure?

Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment.

Which model leads the published Vibe Code Bench snapshot?

Claude Opus 4.8 currently leads the published Vibe Code Bench snapshot with 82.72% vibe code score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on Vibe Code Bench?

47 AI models are included in BenchLM's mirrored Vibe Code Bench snapshot, based on the public leaderboard captured on May 28, 2026.

Last updated: May 28, 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.