PulseBench-Tab
Models convert 1,820 table images in nine languages into HTML tables. T-LAG represents adjacent cells as directed edges and matches their text and structure against human annotations. The reported score is F1 over matched edge weights.
Scores exclude missing outputs, so coverage must accompany comparisons. The maker includes its own product. September 2026 label and identifier corrections do not establish a distinct task version. One Chinese reference table contains placeholder characters.
- Made by
- Pulse Software Corp
- Released
- April 2026
- Kind
- Documents, charts and text in images
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- Many languages
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Many languages
- Strongest model tested
- GPT-6 Astra #4 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 19th, 2026
Pulse Ultra 2 tops the leaderboard with 0.93 Claude Fable 5.1 leads with 0.89 among models released by September 2026
Best score per entry from the benchmark's own leaderboard, read on November 19th, 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. The best model we track is Claude Fable 5.1, at 0.89. 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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- Claude Opus 5.5
- and 13 more
- 0.93 Pulse Ultra 2
- 0.89 Claude Fable 5.1
- 0.89
Gemini 3.8 Flash
- 0.89 GPT-6 Astra
- 0.88 GPT-5.6 Sol
- 0.86 Mistral OCR 4.1
- 0.84 LlamaParse (Agentic)
- 0.82 Extend 2.0
- 0.82 LlamaParse (Agentic Plus)
- 0.81 Reducto r-1
- 0.81 Reducto (Agentic)
- 0.78 LlamaParse (Cost Effective)
- 0.78 Datalab
- 0.76 Azure Document Intelligence
- 0.75 Extend (Lite)
- 0.71 Reducto
- 0.71 Snowflake Document AI
- 0.63 Databricks Document AI
- 1 Pulse Ultra 2 run month 2026-04 0.93
- 2 Claude Fable 5.1 run month 2026-09 0.89
- 3
Gemini 3.8 Flash run month 2026-09 0.89
- 4 GPT-6 Astra run month 2026-09 0.89
- 5 GPT-5.6 Sol run month 2026-09 0.88
- 6 Mistral OCR 4.1 run month 2026-09 0.86
- 7 LlamaParse (Agentic) run month 2026-09 0.84
- 8 Extend 2.0 run month 2026-09 0.82
- 9 LlamaParse (Agentic Plus) run month 2026-09 0.82
- 10 Reducto r-1 run month 2026-09 0.81
- 11 Reducto (Agentic) run month 2026-09 0.81
- 12 LlamaParse (Cost Effective) run month 2026-09 0.78
- 13 Datalab run month 2026-04 0.78
- 14 Azure Document Intelligence run month 2026-04 0.76
- 15 Extend (Lite) run month 2026-09 0.75
- 16 Reducto run month 2026-09 0.71
- 17 Snowflake Document AI run month 2026-04 0.71
- 18 Databricks Document AI run month 2026-04 0.63
Source: Official PulseBench-Tab leaderboard
Source: Official PulseBench-Tab leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Pulse Ultra 2 | 0.93 | run month 2026-04 | Official leaderboard | |
| 2 | Claude Fable 5.1 | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 3 | | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 4 | GPT-6 Astra | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 5 | GPT-5.6 Sol | 0.88 | run month 2026-09 | Jul 2026 | Official leaderboard |
| 6 | Mistral OCR 4.1 | 0.86 | run month 2026-09 | Official leaderboard | |
| 7 | LlamaParse (Agentic) | 0.84 | run month 2026-09 | Official leaderboard | |
| 8 | Extend 2.0 | 0.82 | run month 2026-09 | Official leaderboard | |
| 9 | LlamaParse (Agentic Plus) | 0.82 | run month 2026-09 | Official leaderboard | |
| 10 | Reducto r-1 | 0.81 | run month 2026-09 | Official leaderboard | |
| 11 | Reducto (Agentic) | 0.81 | run month 2026-09 | Official leaderboard | |
| 12 | LlamaParse (Cost Effective) | 0.78 | run month 2026-09 | Official leaderboard | |
| 13 | Datalab | 0.78 | run month 2026-04 | Official leaderboard | |
| 14 | Azure Document Intelligence | 0.76 | run month 2026-04 | Official leaderboard | |
| 15 | Extend (Lite) | 0.75 | run month 2026-09 | Official leaderboard | |
| 16 | Reducto | 0.71 | run month 2026-09 | Official leaderboard | |
| 17 | Snowflake Document AI | 0.71 | run month 2026-04 | Official leaderboard | |
| 18 | Databricks Document AI | 0.63 | run month 2026-04 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 2 | | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 3 | GPT-6 Astra | 0.89 | run month 2026-09 | Sep 2026 | Official leaderboard |
| 4 | GPT-5.6 Sol | 0.88 | run month 2026-09 | Jul 2026 | Official leaderboard |
2 models have held number one on the official leaderboard since its release in April 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official PulseBench-Tab leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official PulseBench-Tab 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 Vision
- MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy.
Qwen 3.6 Plus 86%
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy.
Seed2.1 Pro 90.7%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy.
Qwen 3.8 Max 97.7%
- MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy.
Qwen 3.8 Max 85.9%
Data audit
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (18 entries).
- Pass Models scored 4 models we track, plus 14 other entries on its leaderboard.
- Pass Newest score The newest is from September 2026, for GPT-6 Astra.
- Note Frontier models Only 2 of the top 10 have a score. The strongest tested is GPT-6 Astra (#4).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 4 rows for models we track and 14 other entries at the last read.
- Pass Last read November 19th, 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.