Big Finance Benchmark

Rogo's Big Finance Benchmark evaluates financial-research agents on private, workflow-grounded tasks. Models use Rogo's official agent loop and grading methodology, and the headline measures final-answer accuracy across independent full runs.

Last updated

Main Insights

Results

Big Finance Benchmark results, ranked by score (avg@3), highest first. Last updated .
RankModelProviderScore (avg@3)Cost (USD / task)
1Gemini 4 ArgonGoogle54.7%$0.47
2GPT-6 AstraOpenAI54.6%$0.43
3Muse Spark OpenMeta53.3%$0.34
4GPT-6.1 SolOpenAI52.0%$0.053
5Gemini 4 CarbonGoogle50.1%$0.68
6Gemini 4 HydrogenGoogle49.4%$0.15
7Claude Opus 5Anthropic49.2%$1.83
8Claude Opus 5.5Anthropic48.2%$0.91
9Kimi K3.1 ProMoonshot AI48.2%$0.17
10GPT-6 SolOpenAI47.1%$0.085
11GLM 5.3Z.ai45.8%$0.41
12GPT-5.6 SolOpenAI44.2%$0.23
13Claude Sonnet 5.5Anthropic41.5%$0.26
14Kimi K3Moonshot AI41.3%$0.26
15DeepSeek V4.1 FlashDeepSeek40.6%$0.050
16DeepSeek V4 Pro 0813DeepSeek39.3%$0.14
17Ember-1Fireworks39.0%$0.21
18GLM 5.3 FlashZ.ai37.2%$0.021
19GPT-6 LunaOpenAI35.3%$0.0059