Snorkel Working Memory Test
Models answer questions from a document placed among distractors. The evaluation varies the document position and the length of the input.
The authors compare long-context performance with retrieval.
- Made by
- Snorkel AI
- Released
- July 2024
- Kind
- Finding facts in long inputs
- Level
- School
- Scored
- Graded by a model
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Interaction
- Single turn
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- GPT-4 Turbo Apr 2024 #172 on the capability index
- Last model added
- April 2024
- Scores refreshed
- November 20th, 2026
gpt-3.5-turbo (snapshot unspecified) tops the leaderboard with 97.9% GPT-4 Turbo leads with 87.6% among models released by April 2024
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The authors report accuracy at five levels of document depth or distractor load, so this is their plain mean within each experiment. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. 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 GPT-4 Turbo, at 87.6%. This view shows only models we track, each at its best.
The newest model with a score here came out in April 2024, 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
- 1 gpt-3.5-turbo (snapshot unspecified) Context window used with distractors 97.9%
- 2 mistralai/Mixtral-8x7B-Instruct-v0.1 Context window used with distractors 97.2%
- 3 gemini-1.5-pro-preview-0409 [April 2024 snapshot] Context window used with distractors 94.5%
- 4 GPT-4 Turbo Apr 2024, Context window used with distractors 87.6%
- 5 Claude 3 Opus Context window used with distractors 86.2%
- 6 GPT-4 Turbo (Jan 2024) Context window used with distractors 82.8%
- 7 gpt-3.5-turbo-16k Context window used with distractors 82.1%
- 8 Claude 2.1 Answer document depth 57.9%
Source: Official Snorkel Working Memory Test leaderboard
Source: Official Snorkel Working Memory Test leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | gpt-3.5-turbo (snapshot unspecified) | 97.9% | Context window used with distractors | Official leaderboard | |
| 2 | mistralai/Mixtral-8x7B-Instruct-v0.1 | 97.2% | Context window used with distractors | Official leaderboard | |
| 3 | gemini-1.5-pro-preview-0409 [April 2024 snapshot] | 94.5% | Context window used with distractors | Official leaderboard | |
| 4 | GPT-4 Turbo Apr 2024 | 87.6% | Context window used with distractors | Apr 2024 | Official leaderboard |
| 5 | Claude 3 Opus | 86.2% | Context window used with distractors | Feb 2024 | Official leaderboard |
| 6 | GPT-4 Turbo (Jan 2024) | 82.8% | Context window used with distractors | Jan 2024 | Official leaderboard |
| 7 | gpt-3.5-turbo-16k | 82.1% | Context window used with distractors | Official leaderboard | |
| 8 | Claude 2.1 | 57.9% | Answer document depth | Nov 2023 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-4 Turbo Apr 2024 | 87.6% | Context window used with distractors | Apr 2024 | Official leaderboard |
| 2 | Claude 3 Opus | 86.2% | Context window used with distractors | Feb 2024 | Official leaderboard |
| 3 | GPT-4 Turbo (Jan 2024) | 82.8% | Context window used with distractors | Jan 2024 | Official leaderboard |
| 4 | Claude 2.1 | 57.9% | Answer document depth | Nov 2023 | 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 0%, 25% and 50%. GPT-4 Turbo (Jan 2024) leads 3 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.
0%
4 models we track have a score here.
- 1 GPT-4 Turbo (Jan 2024) ↑2 100%
- 2 GPT-4 Turbo Apr 2024 ↓1 100%
- 3 Claude 3 Opus ↓1 96.6%
- 4 Claude 2.1 93.1%
25%
4 models we track have a score here.
- 1 GPT-4 Turbo Apr 2024 96.6%
- 2 GPT-4 Turbo (Jan 2024) ↑1 79.3%
- 3 Claude 3 Opus ↓1 75.9%
- 4 Claude 2.1 44.8%
50%
4 models we track have a score here.
- 1 GPT-4 Turbo (Jan 2024) ↑2 89.7%
- 2 Claude 3 Opus 89.7%
- 3 GPT-4 Turbo Apr 2024 ↓2 82.8%
- 4 Claude 2.1 51.7%
75%
4 models we track have a score here.
- 1 Claude 3 Opus ↑1 89.7%
- 2 GPT-4 Turbo Apr 2024 ↓1 89.7%
- 3 GPT-4 Turbo (Jan 2024) 86.2%
- 4 Claude 2.1 48.3%
100%
4 models we track have a score here.
- 1 GPT-4 Turbo (Jan 2024) ↑2 100%
- 2 GPT-4 Turbo Apr 2024 ↓1 100%
- 3 Claude 2.1 ↑1 96.6%
- 4 Claude 3 Opus ↓2 79.3%
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: Official Snorkel Working Memory Test leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-4 Turbo Apr 2024 | Context window used with distractors | 87.6% | |||
| Claude 3 Opus | Context window used with distractors | 86.2% | |||
| GPT-4 Turbo (Jan 2024) | Context window used with distractors | 82.8% | |||
| GPT-4 Turbo Apr 2024 | Answer document depth | 82.8% | |||
| GPT-4 Turbo (Jan 2024) | Answer document depth | 82.1% | |||
| Claude 2.1 | Answer document depth | 57.9% | |||
| Claude 2.1 | Context window used with distractors | 53.1% |
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: Official Snorkel Working Memory Test leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-4 Turbo Apr 2024 | Context window used with distractors | 87.6% | |||
| Claude 3 Opus | Context window used with distractors | 86.2% | |||
| GPT-4 Turbo (Jan 2024) | Context window used with distractors | 82.8% | |||
| GPT-4 Turbo Apr 2024 | Answer document depth | 82.8% | |||
| GPT-4 Turbo (Jan 2024) | Answer document depth | 82.1% | |||
| Claude 2.1 | Answer document depth | 57.9% | |||
| Claude 2.1 | Context window used with distractors | 53.1% |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Snorkel Working Memory Test leaderboard
Claude 3 Opus on the official leaderboard passed 85% in February 2024
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.
Source: Official Snorkel Working Memory Test 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
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 official leaderboard (8 entries).
- Pass Models scored 4 models we track, plus 4 other entries on its leaderboard.
- Note Newest score The newest came 2 years ago, in April 2024, for GPT-4 Turbo. 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-4 Turbo (#172).
- Note Headroom Saturated. The best is 87.6%, past 85% since February 2024.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 7 rows for models we track and 4 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.