Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief

Vals Index v2 (Vals Index)

Vals AI composite benchmark across professional finance, coding, modeling, and legal-work tasks, including Finance Agent v2, EMB, Terminal-Bench 2.1, Vibe Code Bench, Code Migration, Legal Research, and HLAB.

Data verified 24 confirmed releases in the last 30 daysStart free brief

How BenchLM shows Vals Index

BenchLM mirrors the public Vals AI Vals Index leaderboard captured from https://www.vals.ai/benchmarks/vals_index and updated by Vals on August 19, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

Vals Index 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.

Snapshot

48 Vals rows8 task viewsprivate datasetTasks: Overall, Finance Agent v2, EMB, Terminal-Bench 2.1, Vibe Code BenchDisplay only

Vals Index score on Vals Index — August 19, 2026

We mirror the published vals index score view for Vals Index. Claude Opus 5 leads the public snapshot at 67.21%, followed by Claude Fable 5 (66.04%) and GPT-5.6 Sol (63.71%). We do not use these results to rank models overall.

48 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated August 19, 2026

Vals Index score table (48 models)

Score
1
Claude Opus 5Anthropic · Closed
67.21%
2
Claude Fable 5Anthropic · Closed
66.04%
3
GPT-5.6 SolOpenAI · Closedmax reasoning
63.71%
4
Claude Opus 4.8Anthropic · Closed
60.91%
5
GPT-5.6 LunaOpenAI · Closedmax reasoning
59.88%
6
Claude Sonnet 5Anthropic · Closed
59.61%
7
GPT-5.6 TerraOpenAI · Closedmax reasoning
59.59%
8
Gemini 3.7 FlashGoogle · Closedhigh reasoning
59.31%
9
Grok 4.6xAI · Closedhigh reasoning
59.17%
10
Kimi K3Moonshot AI · Closed
57.81%
11
GPT-5.5OpenAI · Closedxhigh reasoning
57.41%
12
Muse Spark 1.2Meta · Closedxhigh reasoning
57.05%
13
Claude Opus 4.7Anthropic · Closed
56.11%
14
Gemini 3.6 FlashGoogle · Closedhigh reasoning
55.35%
15
Muse Spark 1.1Meta · Closedxhigh reasoning
54.75%
16
DeepSeek V4 Flash 0731DeepSeek · Closedhigh reasoning
53.57%
17
GLM-5.2Z.AI · Open weight
53.12%
18
Gemini 3.5 FlashGoogle · Closedhigh reasoning
53.08%
19
DeepSeek V4 Pro 0813DeepSeek · Closedmax reasoning
52.37%
20
Qwen3.8 MaxAlibaba · Open weight
51.84%
21
Grok 4.5xAI · Closedhigh reasoning
51.53%
22
Claude Sonnet 4.6Anthropic · Closed
50.59%
23
Qwen3.8-27BAlibaba · Open weightxhigh reasoning
48.48%
24
Qwen3.7 MaxAlibaba · Closed
44.77%
25
Kimi K2.6Moonshot AI · Open weight
43.47%
26
DeepSeek V4 Pro 0813DeepSeek · Closedmax reasoning
42.89%
27
MiniMax M3MiniMax · Open weight
42.72%
28
Gemini 3.1 Pro PreviewGooglehigh reasoning
41.90%
29
MiMo-V2.5-ProXiaomi · Closed
40.97%
30
MiMo-V2.5Xiaomi · Closed
39.91%
31
GPT-5.4 miniOpenAI · Closedxhigh reasoning
39.63%
32
Qwen3.7 PlusAlibaba · Closed
38.65%
33
Gemini 3.5 Flash-LiteGoogle · Closedhigh reasoning
36.71%
34
InklingThinking Machines Lab · Open weight0.99 reasoning
34.10%
35
GPT-5.4 nanoOpenAI · Closedhigh reasoning
33.00%
36
Qwen3.6 PlusAlibaba · Closed
31.98%
37
Inkling-SmallThinking Machines Lab · Open weight0.99 reasoning
31.86%
38
Gemini 3 Flash PreviewGooglehigh reasoning
29.44%
40
26.30%
41
MiniMax M2.7MiniMax · Open weight
24.56%
42
Grok 4.3xAI · Closedhigh reasoning
24.29%
43
Claude Haiku 4.5 ThinkingAnthropic · Closed
22.90%
45
Mistral Medium 3.5Mistral AIhigh reasoning
17.95%
47
Gemini 3.1 Flash Lite PreviewGooglehigh reasoning
15.46%
48
11.49%

The published Vals Index snapshot places Claude Opus 5 first at 67.21%. The third row is 3.50 points behind. The broader top-10 range is 9.40 points, so many of the published results sit in a relatively narrow band.

48 models have been evaluated on Vals Index. The benchmark falls in the External benchmark mirrors category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. Vals Index is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About Vals Index

Year

2026

Tasks

Finance, coding, spreadsheet modeling, code migration, and legal-work components

Format

Composite score

Difficulty

Private economic-work benchmark composite

We mirror Vals Index v2 as a display-only external composite. The task mix changed from the earlier index, so the current score should not be treated as directly interchangeable with an older snapshot. Vals proprietary aggregate scores remain outside weighted rankings.

BenchLM freshness & provenance

Version

Vals Index 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.

FAQ

What does Vals Index measure?

Vals AI composite benchmark across professional finance, coding, modeling, and legal-work tasks, including Finance Agent v2, EMB, Terminal-Bench 2.1, Vibe Code Bench, Code Migration, Legal Research, and HLAB.

Which model leads the published Vals Index snapshot?

Claude Opus 5 currently leads the published Vals Index snapshot with 67.21% vals index score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on Vals Index?

The August 19, 2026 contains 48 AI models.

Last updated: August 19, 2026 · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.