BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

CorpFin

Questions over real corporate finance documents. Run by Vals AI.

As of 25 Oct 2026, Claude Opus 5 leads CorpFin on BenchLeader with 73.2%, ahead of Claude Fable 5 at 71.8%, across 124 model configurations with a published result.

Published by
Vals AI
Category
Knowledge
Index weight
0.5
Models
124
Data as of
25 Oct 2026
Newest result
12 Aug 2026

The publisher has posted no new results on this board since 12 Aug 2026. BenchLeader re-reads it every day, so newer models appear here as soon as they are published.

Vals AI (vals.ai).

What the test looks like

CorpFin asks questions over real corporate finance documents such as credit agreements, requiring the model to find and combine figures across long filings.

How it is scored

Percent of questions answered correctly, run by Vals AI.

What to keep in mind

Document-grounded, so it rewards retrieval within a long context as much as finance knowledge.

Top 15
  1. 1Claude Opus 573.2%
  2. 2Claude Fable 571.8%
  3. 3Kimi K371.6%
  4. 4Muse Spark 1.1 (xhigh)71.3%
  5. 5Muse Spark 1.2 (xhigh)70.9%
  6. 6Inkling Small69.6%
  7. 7Grok 4.368.5%
  8. 8GPT-5.5 (xhigh)68.4%
  9. 9Kimi K2.5 (thinking)68.3%
  10. 10MiniMax M368.1%
  11. 11Qwen3 Max (max)68.0%
  12. 12Grok 4.5 (high)67.4%
  13. 13Claude Opus 4.6 (thinking)67.0%
  14. 14Claude Sonnet 567.0%
  15. 15Grok 4 Fast (thinking)66.9%
124 of 124
#
1Claude Opus 573.2%65.12026-08-12
2Claude Fable 571.8%66.52026-08-12
3Kimi K371.6%58.82026-08-12
4Muse Spark 1.1xhigh71.3%61.62026-08-12
5Muse Spark 1.2xhigh70.9%63.22026-08-12
6Inkling Small69.6%54.92026-08-12
7Grok 4.368.5%54.62026-08-12
8GPT-5.5xhigh68.4%66.12026-08-12
9Kimi K2.5thinking68.3%57.62026-08-12
10MiniMax M368.1%56.42026-08-12
11Qwen3 Maxmax68.0%52.82026-08-12
12Grok 4.5high67.4%59.32026-08-12
13Claude Opus 4.6thinking67.0%62.52026-08-12
14Claude Sonnet 567.0%55.92026-08-12
15Grok 4 Fastthinking66.9%51.42026-08-12
16Kimi K2.666.7%55.32026-08-12
17Claude Opus 4.866.7%60.12026-08-12
18Qwen3.6 Maxmax66.5%62.32026-08-12
19Gemini 3 Flashhigh66.4%58.32026-08-12
20Grok 4.6high66.2%62.92026-08-12
21GLM 5.266.1%51.82026-08-12
22Claude Opus 4.766.1%63.22026-08-12
23Grok 466.0%58.02026-08-12
24Grok 4.1 Fastthinking66.0%54.72026-08-12
25GPT-5.2xhigh65.9%60.72026-08-12
26Qwen3.8 Maxmax65.8%63.32026-08-12
27Nemotron 3 Ultra 550B A55B65.5%55.52026-08-12
28DeepSeek V4 Promax65.4%60.82026-08-12
29GPT-5.6 Terraxhigh65.3%62.82026-08-12
30Qwen3.5 Plusthinking65.3%57.32026-08-12
31Claude Sonnet 4.665.3%57.12026-08-12
32GPT-5.4xhigh65.3%64.72026-08-12
33Muse Spark65.1%64.32026-08-12
34Claude Opus 4.5thinking65.1%59.92026-08-12
35Gemini 3.5 Flashhigh64.7%63.12026-08-12
36Gemini 3.1 Prohigh64.5%60.32026-08-12
37GLM 5.164.5%55.92026-08-12
38GPT-5.6 Solmax64.4%68.02026-08-12
39GPT-5.6 Lunamax64.2%59.92026-08-12
40GPT-5.1high63.8%57.72026-08-12
41Qwen3.7max63.7%60.62026-08-12
42Gemini 3 Prohigh63.7%60.02026-08-12
43Grok 4.20thinking63.7%58.62026-08-12
44Qwen3.5 Flash63.6%52.02026-08-12
45Gemini 3.6 Flashhigh63.3%62.02026-08-12
46GPT-4.1high63.0%46.72026-08-12
47GLM-5thinking62.9%58.92026-08-12
48Qwen3.6 27B62.3%48.32026-08-12
49Claude Sonnet 4.5thinking62.0%52.62026-08-12
50Qwen3.6 Plus61.9%58.82026-08-12
51Ling 3.0 Flash61.9%52.82026-08-12
52DeepSeek V4 Flashhigh61.9%57.32026-08-12
53MiMo-V2.5-Pro61.4%54.92026-08-12
54Claude Opus 4.561.3%57.92026-08-12
55Claude Sonnet 4thinking61.2%53.22026-08-12
56MiniMax M2.761.2%55.62026-08-12
57GPT-5.4 Nanohigh61.2%48.42026-08-12
58Grok 3 Mini Fasthigh61.1%53.72026-08-12
59GPT-5high61.1%57.92026-08-12
60Mistral Large 361.0%45.42026-08-12

More knowledge benchmarks · All LLM benchmarks

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

CorpFin: questions

What does CorpFin measure?
CorpFin asks questions over real corporate finance documents such as credit agreements, requiring the model to find and combine figures across long filings. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads CorpFin?
Claude Opus 5 leads CorpFin with 73.2% as of 25 Oct 2026, ahead of Claude Fable 5 at 71.8%.
How many models have CorpFin results?
124 model configurations have a CorpFin result on BenchLeader, all taken from Vals AI.
Who runs CorpFin and how often is it updated?
CorpFin is published by Vals AI. BenchLeader re-reads the published results once a day and records the date each result was published.
Does CorpFin count toward the BenchLeader Index?
Yes. CorpFin contributes to the knowledge category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.