Benchmark Almanac

SEC filing claim verification (400 items)

Claude Sonnet 5.5 99.5% leads since October 2026
Not in model cards

Models classify whether an excerpt from a corporate filing supports a claim, contradicts it, or provides insufficient evidence.

Vals tested 12 systems on 400 claims from 48 companies and later ran Mercury Decide. The later ClaimBench uses 1,960 claims.

Made by
Vals AI
Released
October 2026
Kind
Faithful to sources
Level
University
Scored
Checked automatically
Answers
Multiple choice
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Items from
Real work and use
Run by
A host
In views
Law and finance
Strongest model tested
GPT-6 Astra #4 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
12
Entries on its own leaderboard, 11 models we track

Claude Sonnet 5.5 leads with 99.5% among models released by September 2026 Claude Sonnet 5.5 leads with 99.5% among models released by September 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 September 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 Claude Sonnet 5.5 99.5%
  2. 2 Claude Opus 5.5 99.3%
  3. 3 Gemini 4 Argon 99%
  4. 4 GPT-5.6 Sol 98.3%
  5. 5 GPT-5.6 Terra 98%
  6. 6 DeepSeek V4.1 Flash 97.8%
  7. 7 GPT-6 Astra 97.5%
  8. 8 GPT-6 Luna 97.5%
  9. 9 GPT-6.1 Sol 97.5%
  10. 10 Jev 1.13.0 97.5%
  11. 11 GPT-5.4 Mini 97.3%
  12. 12 Claude Haiku 4.5 95.3%

Source: Official SEC filing claim verification (400 items) leaderboard

#EntryScoreSettingReleasedRun onSource
1 Claude Sonnet 5.5 99.5% Sep 2026 Oct 2026 Official leaderboard
2 Claude Opus 5.5 99.3% Sep 2026 Oct 2026 Official leaderboard
3 Gemini 4 Argon 99% Sep 2026 Oct 2026 Official leaderboard
4 GPT-5.6 Sol 98.3% Jul 2026 Oct 2026 Official leaderboard
5 GPT-5.6 Terra 98% Jul 2026 Oct 2026 Official leaderboard
6 DeepSeek V4.1 Flash 97.8% Sep 2026 Oct 2026 Official leaderboard
7 GPT-6 Astra 97.5% Sep 2026 Oct 2026 Official leaderboard
8 GPT-6 Luna 97.5% Sep 2026 Oct 2026 Official leaderboard
9 GPT-6.1 Sol 97.5% Sep 2026 Oct 2026 Official leaderboard
10 Jev 1.13.0 97.5% Oct 2026 Official leaderboard
11 GPT-5.4 Mini 97.3% Mar 2026 Oct 2026 Official leaderboard
12 Claude Haiku 4.5 95.3% Oct 2025 Oct 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 Claude Sonnet 5.5 99.5% Sep 2026 Oct 2026 Official leaderboard
2 Claude Opus 5.5 99.3% Sep 2026 Oct 2026 Official leaderboard
3 Gemini 4 Argon 99% Sep 2026 Oct 2026 Official leaderboard
4 GPT-5.6 Sol 98.3% Jul 2026 Oct 2026 Official leaderboard
5 GPT-5.6 Terra 98% Jul 2026 Oct 2026 Official leaderboard
6 DeepSeek V4.1 Flash 97.8% Sep 2026 Oct 2026 Official leaderboard
7 GPT-6 Astra 97.5% Sep 2026 Oct 2026 Official leaderboard
8 GPT-6 Luna 97.5% Sep 2026 Oct 2026 Official leaderboard
9 GPT-6.1 Sol 97.5% Sep 2026 Oct 2026 Official leaderboard
10 GPT-5.4 Mini 97.3% Mar 2026 Oct 2026 Official leaderboard
11 Claude Haiku 4.5 95.3% Oct 2025 Oct 2026 Official leaderboard

Rankings by part

We read 5 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Direct support, Subtle contradiction and Numeric discrepancy. Claude Haiku 4.5 leads 2 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Direct support

11 models we track have a score here.

  1. 1 Claude Haiku 4.5 ↑10 100%
  2. 2 GPT-5.4 Mini ↑8 98.8%
  3. 3 GPT-5.6 Terra ↑2 98.8%
  4. 4 Claude Opus 5.5 ↓2 98.8%
  5. 5 GPT-6 Luna ↑3 98.8%
  6. 6 Claude Sonnet 5.5 ↓5 98.8%
  7. 7 GPT-5.6 Sol ↓3 97.5%
  8. 8 GPT-6 Astra ↓1 97.5%
  9. 9 DeepSeek V4.1 Flash ↓3 97.5%
  10. 10 GPT-6.1 Sol ↓1 97.5%

Accuracy against cost

The study compares accuracy on 400 claims with the cost of evaluating 1,000 claims. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

90% 95% 100% $0.01$0.1$1$10$100 Cost per 1,000 claims ($), log scale, lower is better Accuracy Claude Sonnet 5.5GPT-5.6 SolGPT-5.6 TerraDeepSeek V4.1 FlashJev 1.13.0 90% 95% 100% $0.01$0.1$1$10$100 Cost per 1,000 claims ($), log scale, lower is better Accuracy Claude Sonnet 5.5GPT-5.6 SolGPT-5.6 TerraDeepSeek V4.1 FlashJev 1.13.0
ModelSettingAccuracyCost per 1,000 claims
Claude Sonnet 5.5 99.5% $3.42
Claude Opus 5.5 99.3% $7.28
Gemini 4 Argon 99% $6.64
GPT-5.6 Sol 98.3% $1.85
GPT-5.6 Terra 98% $1.29
DeepSeek V4.1 Flash 97.8% $0.3
Jev 1.13.0 97.5% $0.02
GPT-6 Luna 97.5% $0.08
GPT-6.1 Sol 97.5% $2.71
GPT-6 Astra 97.5% $12.29
GPT-5.4 Mini 97.3% $2.05
Claude Haiku 4.5 95.3% $0.56

Accuracy against cost

The study compares accuracy on 400 claims with the cost of evaluating 1,000 claims. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

90% 95% 100% $0.01$0.1$1$10$100 Cost per 1,000 claims ($), log scale, lower is better Accuracy Claude Sonnet 5.5GPT-5.6 SolGPT-5.6 TerraDeepSeek V4.1 FlashGPT-6 Luna 90% 95% 100% $0.01$0.1$1$10$100 Cost per 1,000 claims ($), log scale, lower is better Accuracy Claude Sonnet 5.5GPT-5.6 SolGPT-5.6 TerraDeepSeek V4.1 FlashGPT-6 Luna
ModelSettingAccuracyCost per 1,000 claims
Claude Sonnet 5.5 99.5% $3.42
Claude Opus 5.5 99.3% $7.28
Gemini 4 Argon 99% $6.64
GPT-5.6 Sol 98.3% $1.85
GPT-5.6 Terra 98% $1.29
DeepSeek V4.1 Flash 97.8% $0.3
GPT-6 Luna 97.5% $0.08
GPT-6.1 Sol 97.5% $2.71
GPT-6 Astra 97.5% $12.29
GPT-5.4 Mini 97.3% $2.05
Claude Haiku 4.5 95.3% $0.56

Number one over time

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

Anthropic
Own leaderboard Passed 85% in Oct 2026, by Claude Sonnet 5.5

Source: Official SEC filing claim verification (400 items) leaderboard

Claude Haiku 4.5 on the official leaderboard passed 85% in October 2025

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
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Sonnet 5.5 GPT-5.4 Mini Claude Haiku 4.5 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Sonnet 5.5 Claude Haiku 4.5

Source: Official SEC filing claim verification (400 items) 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 Factuality

All 164 in Factuality
  • SimpleVQA In use Models answer 2,025 short factual questions about images across 9 tasks and 9 topics. An LLM judge grades answers against fixed references. Qwen3.7-Plus 81.7%
  • HallusionBench In use Models answer 1,129 yes or no questions about 346 images, built in pairs to expose language hallucination and visual illusion. Scored by accuracy. Keye-VL-2.0-30B-A3B 73.5%
  • SimpleQA Verified In use 1,000 human-verified, de-duplicated short-answer factual questions measuring parametric knowledge. Gemini 3.1 Pro 77.5%
  • AA-Omniscience v1.1 In use Models answer 3,000 knowledge questions in six domains without tools and may abstain. The index adds correct answers and subtracts wrong ones. Inkling 2 Small 16.1
  • GaRAGe In use Models write long-form RAG answers to 2,366 questions with human-annotated grounding passages. Scoring checks whether answers use the relevant passages. Amazon Nova Pro 60.7%
  • Kimi Hallu Bench In use Models answer hard prompts drawn from practical work tasks. Fact-checking agents score the share of claims that are hallucinated. GPT-6 Astra 4.20%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (12 entries).
  • Pass Models scored 11 models we track, plus 1 other entry on its leaderboard.
  • Pass Newest score The newest is from September 2026, for Gemini 4 Argon.
  • Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Note Headroom Saturated. The best is 99.5%, past 85% since October 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on vals.ai: 11 rows for models we track and 1 other entry 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