Benchmark Almanac

RAFT (Real-world Annotated Few-shot Tasks)

GPT-3 175B (davinci) 62.7% leads since September 2021
Not in model cards

Models learn text classification tasks from small labeled samples. The tasks use real documents, including customer messages, court decisions, and scientific abstracts.

Made by
Neel Alex and colleagues
Released
September 2021
Kind
Reading comprehension
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
Public, plus a private test set
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Items from
Real work and use
Refresh
One fixed set
Run by
A host
In views
Law and finance
Last model added
March 2021
Scores refreshed
November 20th, 2026
5
Entries in its paper, 3 models we track

GPT-3 175B (davinci) leads with 62.7% among models released by March 2021 GPT-3 175B (davinci) leads with 62.7% among models released by March 2021

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. 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 March 2021, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-3 175B (davinci) Few-shot 62.7%
  2. 2 AdaBoost Supervised 51.4%
  3. 3 GPT-Neo-2.7B Few-shot 48.1%
  4. 4 GPT-2 (1.5B) Few-shot 45.8%
  5. 5 BART MNLI Zero-shot 38.2%

Source: The RAFT (Real-world Annotated Few-shot Tasks) paper

Source: The RAFT (Real-world Annotated Few-shot Tasks) paper

#EntryScoreSettingDateSource
1 GPT-3 175B (davinci) 62.7% Few-shot May 2020 Official leaderboard
2 AdaBoost 51.4% Supervised Official leaderboard
3 GPT-Neo-2.7B 48.1% Few-shot Mar 2021 Official leaderboard
4 GPT-2 (1.5B) 45.8% Few-shot Nov 2019 Official leaderboard
5 BART MNLI 38.2% Zero-shot Official leaderboard
#ModelScoreSettingDateSource
1 GPT-3 175B (davinci) 62.7% Few-shot May 2020 Official leaderboard
2 GPT-Neo-2.7B 48.1% Few-shot Mar 2021 Official leaderboard
3 GPT-2 (1.5B) 45.8% Few-shot Nov 2019 Official leaderboard

Rankings by part

We read 11 parts from the benchmark’s paper, each ranking models on one part of the test, such as ADE Corpus V2, Banking77 and NeurIPS impact statement risks. GPT-3 175B (davinci) leads 10 of them, and GPT-Neo-2.7B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

ADE Corpus V2

3 models we track have a score here.

  1. 1 GPT-3 175B (davinci) 68.6%
  2. 2 GPT-2 (1.5B) ↑1 60%
  3. 3 GPT-Neo-2.7B ↓1 45.2%

Number one over time

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

OpenAI
Benchmark paper Not tested since Sep 2021

Source: The RAFT (Real-world Annotated Few-shot Tasks) paper

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
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2020 2021 Near the ceiling (85%) GPT-3 175B (davinci) GPT-2 (1.5B) 0% 20% 40% 60% 80% 100% 2020 2021 Near the ceiling (85%) GPT-3 175B (davinci) GPT-2 (1.5B)

Source: Official RAFT (Real-world Annotated Few-shot Tasks) 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 Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 5 years ago, in March 2021, for GPT-Neo-2.7B. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (5 entries).
  • Pass Models scored 3 models we track, plus 2 other entries on its leaderboard.
  • Warning Newest score The newest came 5 years ago, in March 2021, for GPT-Neo-2.7B.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 62.7%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 4 rows for models we track and 2 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