Industry3 benchmarks

Investment Banking

Financial analysis over filings and credit agreements — valuation math, document QA, and the quantitative reasoning behind deals.

The Investment Bankingscore is the mean of a model’s normalized 0–100 scores (direction-aware, so lower-is-better metrics are inverted) across the 3 benchmarks below — the same figure the leaderboard’s industry view ranks by.

Leaders

DeepSeek V4-Flash leads this industry with a score of 87.2.

DeepSeek V4-Flash87.2

Nemotron 3 Ultra86.9

Qwen3.6-35B-A3B85.6

Nova 2 Pro81.5

GPT-5.581.0

Claude Opus 4.780.2

Claude Opus 4.880.2

DeepSeek V4-Pro79.7

Gemini 3.1 Pro79.4

Claude Opus 4.679.2

GPT-5.279.2

Qwen3.6-27B78.8

Kimi K2.678.6

Gemini 3 Flash78.4

Qwen3.7 Max78.1

Gemini 3 Pro77.8

Kimi K2 Thinking76.6

GLM-5.175.4

Gemini 2.5 Pro73.6

DeepSeek V3.272.8

Grok 4.368.5

Llama 4 Maverick66.7

Claude Sonnet 4.665.3

GPT-5.465.3

Muse Spark65.1

Gemini 3.5 Flash64.7

Grok 4.2063.7

Claude Haiku 4.560.6

Llama 4 Scout59.4

Mistral Large 352.5

Benchmarks in this score

Each model’s scores on these are normalized and averaged to produce the industry score above.

Finance%

Corporate Finance

Vals AI CorpFin v2 — expert-built questions over long-context corporate credit agreements; an independent, in-house-run finance benchmark.

Knowledge%

MMLU-Pro

A harder, cleaned-up successor to MMLU spanning 57+ subjects with 10-way multiple choice and reasoning-heavy items.

Reasoning%

GPQA Diamond

Graduate-level, Google-proof science questions (physics, chemistry, biology) written by domain experts to resist web lookup.