AI Models on Finance Benchmarks

AI models scored on public finance benchmarks (analyst research, Excel modeling, corporate tax), with sample sizes, margins and the price per million tokens on the date we checked.

Scores as published by Vals AI, sorted by Finance Agent v2 Context: FAB v2: n=450 test questions, 3 runs · Oct 12, 2026 leaderboard · Vals AI · ± SEM Context: EMB: n=51 test tasks · Oct 12, 2026 leaderboard · Vals AI · ± as published, type not stated Context: Tax Agent: n=193 test questions · Oct 12, 2026 leaderboard · Vals AI · ± as published, type not stated
Model	Maker	FAB v2	EMB	Tax Agent	API price per 1M tokens (in / out)
Gemini 4 Argon	Google	65.40% ±0.32	75.24% ±2.46	44.56% ±3.59	$4.00 / $20.00
Gemini 3.8 Flash	Google	61.44% ±0.13	72.20% ±2.42	32.12% ±3.37	$1.50 / $7.50
Muse Spark 1.2	Meta	60.60% ±0.28	56.98% ±3.13	32.80% ±2.75	$1.25 / $4.25
Claude Opus 5.5	Anthropic	58.59% ±0.17	75.94% ±2.38	45.57% ±3.59	$4.00 / $20.00
GLM 5.3 Flash	Z.ai (zAI)	57.85% ±2.06	55.93% ±3.56	31.55% ±3.35	$0.07 / $0.25
GPT-5.6 Luna	OpenAI	55.04% ±0.31	67.12% ±2.94	24.82% ±3.11	$0.20 / $1.20
Ling 3.0 Flash Fin	Ant Group	54.93% ±0.68	36.83% ±3.33	20.60% ±2.91	$0.06 / $0.18
Grok 4.6	SpaceXAI	53.68% ±0.67	62.73% ±3.08	33.14% ±3.40	$2.00 / $6.00

A benchmark is not a trading test. These scores measure answers to analyst, modeling and tax questions graded by AI judges. Differences smaller than the published margins are not meaningful.

Head-to-head comparisons

We only publish a comparison when both models have results on the same published benchmarks. See also the AI Analyst Scoreboard and the AI tools directory.