BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

PRBench Finance

Professional reasoning in finance, graded by practitioners. Scale AI.

As of 25 Oct 2026, Muse Spark 1.3 leads PRBench Finance on BenchLeader with 59.5%, ahead of Muse Wonder at 58.7%, across 37 model configurations with a published result.

Published by
Scale AI SEAL
Category
Knowledge
Index weight
0.5
Models
37
Data as of
25 Oct 2026
Newest result
16 Oct 2026

Scale AI SEAL Leaderboards.

What the test looks like

PRBench Finance poses open-ended professional questions such as valuing a business or assessing a filing, and practitioners grade the answers against rubrics.

How it is scored

Percent of rubric points achieved, published by Scale AI.

What to keep in mind

Human rubric grading and a modest task count; scores reflect judgment and presentation as well as correctness.

Top 15
  1. 1Muse Spark 1.359.5%
  2. 2Muse Wonder58.7%
  3. 3Muse Spark 1.155.0%
  4. 4Claude Fable 553.9%
  5. 5Claude Opus 4.6 (none)53.3%
  6. 6Muse Spark52.4%
  7. 7GPT-551.3%
  8. 8GPT-5 Pro51.1%
  9. 9Claude Fable 5.150.8%
  10. 10GPT-5.6 Sol (max)50.5%
  11. 11Gemini 4 Argon49.9%
  12. 12Gemini 3.8 Flash49.5%
  13. 13o3-pro49.1%
  14. 14GPT-5.1 (thinking)48.0%
  15. 15o347.7%
37 of 37
#
1Muse Spark 1.359.5%59.82026-09-14
2Muse Wonder58.7%64.32026-10-14
3Muse Spark 1.155.0%62.52026-07-09
4Claude Fable 553.9%66.52026-07-28
5Claude Opus 4.6no reasoning53.3%56.52026-02-17
6Muse Spark52.4%64.32026-04-08
7GPT-551.3%56.22025-11-13
8GPT-5 Pro51.1%59.82025-11-13
9Claude Fable 5.150.8%66.42026-09-03
10GPT-5.6 Solmax50.5%68.02026-07-28
11Gemini 4 Argon49.9%63.62026-10-16
12Gemini 3.8 Flash49.5%55.72026-09-09
13o3-pro49.1%53.42025-11-13
14GPT-5.1thinking48.0%54.12025-12-11
15o347.7%59.52025-11-13
16GPT-6 Astra47.5%68.02026-09-09
17Kimi K2.546.5%54.52026-02-13
18GPT-5.2 Pro46.3%–2025-12-16
19GPT-6.1 Sol46.2%63.02026-10-07
20Claude Opus 4.5thinking46.2%59.92025-11-26
21GPT-5.4high45.6%58.92026-04-22
22gpt-oss-120b43.8%45.62025-11-13
23Claude Sonnet 4.543.8%53.72025-11-13
24Kimi K2 Thinkingthinking43.4%52.42025-11-13
25Gemini 3.1 Pro41.9%60.52026-03-23
26Mistral Medium 3.5medium39.4%41.12025-11-13
27o4-mini39.2%56.92025-11-13
28Gemini 3 Pro39.2%60.42025-11-20
29Qwen3 235B A22B 250739.1%50.42025-11-13
30Gemini 2.5 Pro38.9%53.92025-11-13
31Gemini 2.5 Flash38.4%52.72025-11-13
32Kimi K238.3%50.42025-11-13
33Claude Opus 4.135.1%51.52025-11-13
34GPT-4.134.3%48.32025-11-13
35DeepSeek R1 052832.7%49.52025-11-13
36GPT-4.1 mini30.4%45.02025-11-13
37Llama 4 Maverick Basic22.4%39.42025-11-13

More knowledge benchmarks · All LLM benchmarks

Cite as: BenchLeader, “PRBench Finance leaderboard”, data as of 25 Oct 2026, https://www.benchleader.com/benchmarks/scale_prbench_finance.

PRBench Finance: questions

What does PRBench Finance measure?
PRBench Finance poses open-ended professional questions such as valuing a business or assessing a filing, and practitioners grade the answers against rubrics. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads PRBench Finance?
Muse Spark 1.3 leads PRBench Finance with 59.5% as of 25 Oct 2026, ahead of Muse Wonder at 58.7%.
How many models have PRBench Finance results?
37 model configurations have a PRBench Finance result on BenchLeader, all taken from Scale AI SEAL.
Who runs PRBench Finance and how often is it updated?
PRBench Finance is published by Scale AI SEAL. BenchLeader re-reads the published results once a day and records the date each result was published.
Does PRBench Finance count toward the BenchLeader Index?
Yes. PRBench Finance contributes to the knowledge category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.