AnomalyXL
Models locate and describe anomalies in long synthetic time series. The benchmark checks multiple-choice answers and scores precise answers using the known anomaly properties.
On September 6, 2026, GPT-5.5 with image-based TimeRLM led localization at 0.682 intersection over union. The paper also compares GPT-4.1 mini, Qwen3.5-4B and specialized time series models.
- Made by
- Nicolas Zumarraga and the TimeRLM team
- Released
- August 2026
- Kind
- Reasoning over long documents
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- English
- In and out
- Text, Time series, Images → Text
- Interaction
- Single turn
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision
- Strongest model tested
- GPT-5.5 #21 on the capability index
- Last model added
- April 2026
- Scores refreshed
- November 20th, 2026
TimeRLM image + GPT-5.5 tops the leaderboard with 0.68 GPT-5.5 leads with 0.33 among models released by April 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 reports no overall score. The headline is localization IoU, the first measure in the precise task suite. Where a model ran on its own with several settings, we keep its best. Every run is in the next chart. Agents other companies built around a model get a pale shade of that model's lab color. Gray marks blends and entries not in our model list, such as fine-tunes and older models. None of these count toward leaders. The best model on its own is GPT-5.5, at 0.33. Agents built around a model, such as TimeRLM image + GPT-5.5, are their own entries. We don't count them as the model's score. This view shows only models we track, each at its best.
The newest model with a score here came out in April 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
- GPT-6 Astra
- and 16 more
- 0.68 TimeRLM image + GPT-5.5
- 0.68 TimeRLM text + GPT-5.5
- 0.54 TimeRLM text + Qwen3.5-4B
- 0.37 TimeRLM image + GPT-4.1 mini
- 0.33 GPT-5.5
- 0.32 Toto Qwen3-VL 32B
- 0.27 TimeRLM text + GPT-4.1 mini
- 0.21 TimeRLM image + Qwen3.5-4B
- 0.16 Features + GBT heads
- 0.14 GPT-4.1 mini
- 0.10
Qwen3.5-4B
- 0.01 OpenTSLM Qwen3.5-4B
- 0.00 ITFormer Qwen3.5-4B
- 0.00 ChatTS Qwen3.5-4B
- 1 TimeRLM image + GPT-5.5 0.68
- 2 TimeRLM text + GPT-5.5 0.68
- 3 TimeRLM text + Qwen3.5-4B RL 0.54
- 4 TimeRLM image + GPT-4.1 mini 0.37
- 5 GPT-5.5 image 0.33
- 6 Toto Qwen3-VL 32B 0.32
- 7 TimeRLM text + GPT-4.1 mini 0.27
- 8 TimeRLM image + Qwen3.5-4B 0.21
- 9 Features + GBT heads 0.16
- 10 GPT-4.1 mini image 0.14
- 11
Qwen3.5-4B image 0.10
- 12 OpenTSLM Qwen3.5-4B 0.01
- 13 ITFormer Qwen3.5-4B 0.00
- 14 ChatTS Qwen3.5-4B 0.00
Source: The AnomalyXL paper
- 1 GPT-5.5 image 0.33
- 2 GPT-4.1 mini image 0.14
- 3
Qwen3.5-4B image 0.10
Source: The AnomalyXL paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | TimeRLM image + GPT-5.5 | 0.68 | Apr 2026 | Official leaderboard | |
| 2 | TimeRLM text + GPT-5.5 | 0.68 | Apr 2026 | Official leaderboard | |
| 3 | TimeRLM text + Qwen3.5-4B | 0.54 | RL | Feb 2026 | Official leaderboard |
| 4 | TimeRLM image + GPT-4.1 mini | 0.37 | Apr 2025 | Official leaderboard | |
| 5 | GPT-5.5 | 0.33 | image | Apr 2026 | Official leaderboard |
| 6 | Toto Qwen3-VL 32B | 0.32 | Official leaderboard | ||
| 7 | TimeRLM text + GPT-4.1 mini | 0.27 | Apr 2025 | Official leaderboard | |
| 8 | TimeRLM image + Qwen3.5-4B | 0.21 | Feb 2026 | Official leaderboard | |
| 9 | Features + GBT heads | 0.16 | Official leaderboard | ||
| 10 | GPT-4.1 mini | 0.14 | image | Apr 2025 | Official leaderboard |
| 11 | | 0.10 | image | Feb 2026 | Official leaderboard |
| 12 | OpenTSLM Qwen3.5-4B | 0.01 | Official leaderboard | ||
| 13 | ITFormer Qwen3.5-4B | 0.00 | Official leaderboard | ||
| 14 | ChatTS Qwen3.5-4B | 0.00 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 0.33 | image | Apr 2026 | Official leaderboard |
| 2 | GPT-4.1 mini | 0.14 | image | Apr 2025 | Official leaderboard |
| 3 | | 0.10 | image | Feb 2026 | Official leaderboard |
Rankings by part
We read 11 parts from the benchmark’s paper in 2 groups: anomalyxl-mcq and anomalyxl-localize. Each ranks models on one part of the test. GPT-5.5 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The AnomalyXL paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| TimeRLM image + GPT-5.5 | TimeRLM image | 0.68 | |||
| TimeRLM text + GPT-5.5 | TimeRLM text | 0.68 | |||
| TimeRLM text + GPT-5.5 | depth-1 | TimeRLM text | 0.66 | ||
| TimeRLM text + Qwen3.5-4B | RL | TimeRLM text | 0.54 | ||
| TimeRLM image + GPT-4.1 mini | TimeRLM image | 0.37 | |||
| GPT-5.5 | image | 0.33 | |||
| TimeRLM text + GPT-4.1 mini | TimeRLM text | 0.27 | |||
| TimeRLM image + Qwen3.5-4B | TimeRLM image | 0.21 | |||
| TimeRLM text + Qwen3.5-4B | TimeRLM text | 0.20 | |||
| GPT-4.1 mini | image | 0.14 | |||
| image | 0.10 | ||||
| GPT-5.5 | text | 0.10 | |||
| text | 0.04 | ||||
| GPT-4.1 mini | text | 0.01 |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: The AnomalyXL paper
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.5 | image | 0.33 | |||
| GPT-4.1 mini | image | 0.14 | |||
| image | 0.10 | ||||
| GPT-5.5 | text | 0.10 | |||
| text | 0.04 | ||||
| GPT-4.1 mini | text | 0.01 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The AnomalyXL paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AnomalyXL 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 Long context
- MRCR v2 In use Multi-round coreference: retrieve the n-th of several similar requests hidden in a long synthetic conversation, up to 1M tokens.
Muse Spark 1.3 98.5%
- LongBench v2 In use 503 multiple-choice questions over 8K-2M-word contexts (QA, dialogue, code repos, structured data). Accuracy.
Inkling 2 Small 70.6%
- GraphWalks In use A graph is given as an edge list in long context. The model answers BFS or parent queries, scored by F1. Claude Fable 5 99.96%
- OpenAI MRCR v2 In use Models read a long synthetic chat that hides 2, 4 or 8 identical writing requests and must reproduce the response to the i-th one. Scored by mean string match ratio per context bin.
Gemini 3.7 Flash 89%
- CL-bench In use Model must learn new knowledge from a supplied long context (rules, procedures, data) and solve tasks. Strict LLM-verified rubrics. GPT-5.4 27.9%
- AA-LCR 1.1 In use Models answer 100 questions that need reasoning across sets of real documents of about 100K tokens. An LLM equality checker with a fixed system prompt compares each answer with the key.
Step 5 Preview 88.3%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 7 months ago, in April 2026, for GPT-5.5. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (14 entries).
- Pass Models scored 3 models we track, plus 11 other entries on its leaderboard.
- Warning Newest score The newest came 7 months ago, in April 2026, for GPT-5.5.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
- Note Headroom Scored as a plain number, with no fixed ceiling.
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.