Skip to content

Benchmark

FinBen ConvFinQA EmAcc

Scores are percentages. Higher is better.

Models measured
18
Scores recorded
18
Sources
1
IQ weight
0.45

Score distribution

13.1%
Mean
0%
Median
23.2%
Std dev
0%
10th percentile
43%
90th percentile
0%
Min
0%76%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.45 for chat models, measured over 18 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-4 Turbopublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results76%
2GPT-3.5 Turbopublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results60%
3gemini-1.0-propublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results43%
4Mixtral 8X7b V0.1published-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results31%
5Llama 2 70bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results25%
6Qwen 2 (72B)published-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results1%
7code-llama-7b-instructpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
8codellama-7bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
9DeepSeek R1 Distill Qwen 1.5Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
10DeepSeek R1 Distill Qwen 7Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
11DeepSeek V3published-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
12Llama 3.1 70Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
13Llama 3.1 8Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
14Llama 3.2 3Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
15Llama-2-7b-chat-hfpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
16Qwen2 7B Instructpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
17Qwen2.5 7Bpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%
18Qwen2.5 7B Instructpublished-demo-leaderboard, ConvFinQA, EmAcc· TheFinAI public numeric results0%

Data credits: The Fin AI. Published FinBen result summaries and Open Greek Financial LLM Leaderboard aggregates. Results licence not separately stated; fractional FinBen success metrics converted to percentages.