Benchmark Almanac

MM-FinEval

GPT-5.5 88% leads since September 2026
Not in model cards

Models forecast financial outcomes from earnings call transcripts, slides, and audio. The authors compare 19 models using directional accuracy and absolute error.

Released
September 2026
Kind
Finance
Level
University
Scored
Checked automatically
Answers
Prediction
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Documents, Audio → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
May 2026
Scores refreshed
November 20th, 2026
19
Entries in its paper, 14 models we track

GPT-5.5 leads with 88% among models released by May 2026 GPT-5.5 leads with 88% among models released by May 2026

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper gives no overall directional accuracy, so this is the plain mean of its 12 task accuracies. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in May 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.5 Image-Text 88%
  2. 2 Claude Sonnet 4.6 Image-Text 87.9%
  3. 3 Claude Opus 4.8 Image-Text 87.3%
  4. 4 GPT-5.4 Image-Text 85.8%
  5. 5 Qwen3.5-35B-A3B Image-Text 83.9%
  6. 6 Qwen3.5-27B Image-Text 83.8%
  7. 7 Qwen3.6 27B Image-Text 83.8%
  8. 8 Mis3-14B Image-Text 83.6%
  9. 9 Gemma 4 E2B Any-to-Any 83.4%
  10. 10 Gm3n Image-Text 83%
  11. 11 Mis3-8B Image-Text 81.5%
  12. 12 Qwen 3.6 35B-A3B Image-Text 81%
  13. 13 Qwen3.5-9B Image-Text 80.7%
  14. 14 Gemma 4 E4B Any-to-Any 73.2%
  15. 15 Qwen3.5-4B Image-Text 69.3%
  16. 16 Moss Audio-Text 49.6%
  17. 17 Mis3-3B Image-Text 44.3%
  18. 18 Voxtral Small Audio-Text 34%
  19. 19 Qwen3.5-2B Image-Text 28.2%

Source: The MM-FinEval paper

#EntryScoreSettingDateSource
1 GPT-5.5 88% Image-Text Apr 2026 Official leaderboard
2 Claude Sonnet 4.6 87.9% Image-Text Feb 2026 Official leaderboard
3 Claude Opus 4.8 87.3% Image-Text May 2026 Official leaderboard
4 GPT-5.4 85.8% Image-Text Mar 2026 Official leaderboard
5 Qwen3.5-35B-A3B 83.9% Image-Text Feb 2026 Official leaderboard
6 Qwen3.5-27B 83.8% Image-Text Feb 2026 Official leaderboard
7 Qwen3.6 27B 83.8% Image-Text Apr 2026 Official leaderboard
8 Mis3-14B 83.6% Image-Text Official leaderboard
9 Gemma 4 E2B 83.4% Any-to-Any Mar 2026 Official leaderboard
10 Gm3n 83% Image-Text Official leaderboard
11 Mis3-8B 81.5% Image-Text Official leaderboard
12 Qwen 3.6 35B-A3B 81% Image-Text Apr 2026 Official leaderboard
13 Qwen3.5-9B 80.7% Image-Text Feb 2026 Official leaderboard
14 Gemma 4 E4B 73.2% Any-to-Any Mar 2026 Official leaderboard
15 Qwen3.5-4B 69.3% Image-Text Feb 2026 Official leaderboard
16 Moss 49.6% Audio-Text Official leaderboard
17 Mis3-3B 44.3% Image-Text Official leaderboard
18 Voxtral Small 34% Audio-Text Jul 2025 Official leaderboard
19 Qwen3.5-2B 28.2% Image-Text Feb 2026 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.5 88% Image-Text Apr 2026 Official leaderboard
2 Claude Sonnet 4.6 87.9% Image-Text Feb 2026 Official leaderboard
3 Claude Opus 4.8 87.3% Image-Text May 2026 Official leaderboard
4 GPT-5.4 85.8% Image-Text Mar 2026 Official leaderboard
5 Qwen3.5-35B-A3B 83.9% Image-Text Feb 2026 Official leaderboard
6 Qwen3.5-27B 83.8% Image-Text Feb 2026 Official leaderboard
7 Qwen3.6 27B 83.8% Image-Text Apr 2026 Official leaderboard
8 Gemma 4 E2B 83.4% Any-to-Any Mar 2026 Official leaderboard
9 Qwen 3.6 35B-A3B 81% Image-Text Apr 2026 Official leaderboard
10 Qwen3.5-9B 80.7% Image-Text Feb 2026 Official leaderboard
11 Gemma 4 E4B 73.2% Any-to-Any Mar 2026 Official leaderboard
12 Qwen3.5-4B 69.3% Image-Text Feb 2026 Official leaderboard
13 Voxtral Small 34% Audio-Text Jul 2025 Official leaderboard
14 Qwen3.5-2B 28.2% Image-Text Feb 2026 Official leaderboard

Rankings by part

We read 24 parts from the benchmark’s paper in 2 groups: directional accuracy and mean absolute error. Each ranks models on one part of the test. Claude Sonnet 4.6 leads 7 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

car

14 models we track have a score here.

  1. 1 Gemma 4 E2B ↑7 55.2%
  2. 2 Claude Sonnet 4.6 53.7%
  3. 3 GPT-5.5 ↓2 53.5%
  4. 4 Gemma 4 E4B ↑7 53.2%
  5. 5 Qwen3.5-35B-A3B 53.1%
  6. 6 GPT-5.4 ↓2 51.8%
  7. 7 Qwen3.6 27B 51.6%
  8. 8 Qwen 3.6 35B-A3B ↑1 51.1%
  9. 9 Qwen3.5-27B ↓3 50.6%
  10. 10 Qwen3.5-9B 48.1%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Benchmark paper Passed 85% in Sep 2026, by GPT-5.5
GPT-5.5 88%

Source: The MM-FinEval paper

Claude Sonnet 4.6 on the official leaderboard passed 85% in February 2026

Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.5 Claude Sonnet 4.6 Voxtral Small 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.5 Voxtral Small

Source: Official MM-FinEval leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 637 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
  • Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (19 entries).
  • Pass Models scored 14 models we track, plus 5 other entries on its leaderboard.
  • Note Newest score The newest came 6 months ago, in May 2026, for Claude Opus 4.8. That’s expected for a saturated test.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.5 (#21).
  • Note Headroom Saturated. The best is 88%, past 85% since February 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 14 rows for models we track and 5 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check