Benchmark Almanac

OOLONG

Gemini 3 Pro 68.3 leads since November 2025
Not in model cards

Models aggregate information across long collections of text. The hub port implements the synthetic and real datasets with a context-controlled recursive language model interface.

On October 5, 2026, Gemini 3 Pro averaged 78.69 on OOLONG-synth and 57.93 on OOLONG-real. GPT-5 averaged 70.75 and 47.00, respectively.

Made by
OOLONG authors
Released
September 2025
Kind
Reasoning over long documents
Level
University
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Agentic
Environment
Code sandbox
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
April 2026
Scores refreshed
November 20th, 2026
20
Entries on its own leaderboard, 17 models we track

Gemini 3 Pro leads with 68.3 among models released by April 2026 Gemini 3 Pro leads with 68.3 among models released by April 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The authors average OOLONG-synth and OOLONG-real, with unsupported context lengths counting as zero in each split. 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 April 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 Gemini 3 Pro 68.3
  2. 2 GPT-5 58.9
  3. 3 Gemini-2.5-Pro (snapshot unspecified) 53.7
  4. 4 o3 49.5
  5. 5 GPT-5 mini 49.1
  6. 6 Claude Sonnet 4 47.5
  7. 7 o4-mini 41.9
  8. 8 Gemma 4 31B IT 41.8
  9. 9 GPT-5 nano 40.9
  10. 10 GPT-5.2 39.4
  11. 11 qwen3-4b-instruct-2507 31.8
  12. 12 Olmo 3.1 32B Think 22.9
  13. 13 DeepSeek-R1 May 2025 22.6
  14. 14 Olmo-3-Think-7B 19.4
  15. 15 Olmo 3.1 32B Instruct 18.5
  16. 16 Olmo-3-Instruct-7B 14.5
  17. 17 LLaMA-33B 12.8
  18. 18 Llama 3-8B 12.1
  19. 19 Qwen3-4B thinking 10.1
  20. 20 Llama 4 Maverick 9.22

Source: Official OOLONG leaderboard

#EntryScoreSettingDateSource
1 Gemini 3 Pro 68.3 Nov 2025 Official leaderboard
2 GPT-5 58.9 Aug 2025 Official leaderboard
3 Gemini-2.5-Pro (snapshot unspecified) closed weights 53.7 Official leaderboard
4 o3 49.5 Apr 2025 Official leaderboard
5 GPT-5 mini 49.1 Aug 2025 Official leaderboard
6 Claude Sonnet 4 47.5 May 2025 Official leaderboard
7 o4-mini 41.9 Apr 2025 Official leaderboard
8 Gemma 4 31B IT 41.8 Apr 2026 Official leaderboard
9 GPT-5 nano 40.9 Aug 2025 Official leaderboard
10 GPT-5.2 39.4 Dec 2025 Official leaderboard
11 qwen3-4b-instruct-2507 31.8 Aug 2025 Official leaderboard
12 Olmo 3.1 32B Think 22.9 Dec 2025 Official leaderboard
13 DeepSeek-R1 May 2025 22.6 May 2025 Official leaderboard
14 Olmo-3-Think-7B open weights 19.4 Official leaderboard
15 Olmo 3.1 32B Instruct 18.5 Dec 2025 Official leaderboard
16 Olmo-3-Instruct-7B open weights 14.5 Official leaderboard
17 LLaMA-33B 12.8 Feb 2023 Official leaderboard
18 Llama 3-8B 12.1 Apr 2024 Official leaderboard
19 Qwen3-4B 10.1 thinking Apr 2025 Official leaderboard
20 Llama 4 Maverick 9.22 Apr 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3 Pro 68.3 Nov 2025 Official leaderboard
2 GPT-5 58.9 Aug 2025 Official leaderboard
3 o3 49.5 Apr 2025 Official leaderboard
4 GPT-5 mini 49.1 Aug 2025 Official leaderboard
5 Claude Sonnet 4 47.5 May 2025 Official leaderboard
6 o4-mini 41.9 Apr 2025 Official leaderboard
7 Gemma 4 31B IT 41.8 Apr 2026 Official leaderboard
8 GPT-5 nano 40.9 Aug 2025 Official leaderboard
9 GPT-5.2 39.4 Dec 2025 Official leaderboard
10 qwen3-4b-instruct-2507 31.8 Aug 2025 Official leaderboard
11 Olmo 3.1 32B Think 22.9 Dec 2025 Official leaderboard
12 DeepSeek-R1 May 2025 22.6 May 2025 Official leaderboard
13 Olmo 3.1 32B Instruct 18.5 Dec 2025 Official leaderboard
14 LLaMA-33B 12.8 Feb 2023 Official leaderboard
15 Llama 3-8B 12.1 Apr 2024 Official leaderboard
16 Qwen3-4B 10.1 thinking Apr 2025 Official leaderboard
17 Llama 4 Maverick 9.22 Apr 2025 Official leaderboard

Rankings by part

We read 22 parts from the benchmark’s own leaderboard in 3 groups: dataset, oolong-synth and oolong-real. Each ranks models on one part of the test. Gemini 3 Pro leads 20 of them, and Claude Sonnet 4 leads the other 2. Arrows show how far a model moved from its overall rank among the models we track.

OOLONG-synth

17 models we track have a score here.

  1. 1 Gemini 3 Pro 78.7
  2. 2 GPT-5 70.7
  3. 3 GPT-5 mini ↑1 63.7
  4. 4 o3 ↓1 62.4
  5. 5 Claude Sonnet 4 58.2
  6. 6 o4-mini 56.7
  7. 7 Gemma 4 31B IT 53.4
  8. 8 GPT-5 nano 50.7
  9. 9 GPT-5.2 44.3
  10. 10 qwen3-4b-instruct-2507 40.7

2 models have held number one on the official leaderboard since its release in September 2025

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

OpenAI Google DeepMind
Own leaderboard Not tested since Apr 2026

Source: Official OOLONG leaderboard

Scores over time

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
0.00 20.0 40.0 60.0 80.0 2023 2024 2025 2026 0.00 20.0 40.0 60.0 80.0 2023 2024 2025 2026

Source: Official OOLONG 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

All 115 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 Gemma 4 31B IT. 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 official leaderboard (20 entries).
  • Pass Models scored 17 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 7 months ago, in April 2026, for Gemma 4 31B IT.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on oolongbench.github.io: 17 rows for models we track and 3 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