GDPval-AA
Claude Fable 5 1,932 leads since Jun 2026 Fading
Agents complete GDPval occupational tasks. Artificial Analysis compares the outputs in pairs and ranks models by Elo.
Versions 2 and 2.1 changed the task set, so their Elo scores have their own records.
Made by Artificial Analysis Released Dec 2025 Tasks from real jobs All tasks public Last in a launch post Jun 2026 , dropped for a newer version
Last model added Jun 2026
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Graded by a model A language model scores each response, usually against a rubric.
- Answers
- Open-ended Free-form text, documents, designs or media judged as a whole.
16
Models scored in lab cards
14
Launch posts list it, from 8 labs
14
System cards report it, from 6 labs
Claude Fable 5 leads with 1,932 among models released by Jun 2026
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Jun 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 14 more
- 1,932 Claude Fable 5
- 1,890 Claude Opus 4.8
- 1,753 Claude Opus 4.7
- 1,656
Gemini 3.5 Flash - 1,633 Claude Sonnet 4.6
- 1,606 Claude Opus 4.6
- 1,581 MiMo-V2.5-Pro
- 1,554
DeepSeek-V4-Pro - 1,495
MiniMax-M2.7 - 1,444
Muse Spark - 1,426 MiMo-V2-Pro
- 1,416
Step 3.7 Flash - 1,410 MiMo-V2-Omni
- 1,409 GLM-5
- 1,395 DeepSeek-V4-Flash
- 1,317 Gemini 3.1 Pro
Bars start at 1,200.
- 1 Claude Fable 5 1,932
- 2 Claude Opus 4.8 1,890
- 3 Claude Opus 4.7 1,753
- 4 Gemini 3.5 Flash 1,656
- 5 Claude Sonnet 4.6 1,633
- 6 Claude Opus 4.6 1,606
- 7 MiMo-V2.5-Pro 1,581
- 8 DeepSeek-V4-Pro 1,554
- 9 MiniMax-M2.7 1,495
- 10 Muse Spark 1,444
- 11 MiMo-V2-Pro 1,426
- 12 Step 3.7 Flash 1,416
- 13 MiMo-V2-Omni 1,410
- 14 GLM-5 1,409
- 15 DeepSeek-V4-Flash 1,395
- 16 Gemini 3.1 Pro 1,317
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Fable 5 | 1,932 | Jun 2026 | Model card | |
| 2 | Claude Opus 4.8 | 1,890 | May 2026 | Model card | |
| 3 | Claude Opus 4.7 | 1,753 | Apr 2026 | Model card | |
| 4 | Gemini 3.5 Flash | 1,656 | May 2026 | Model card | |
| 5 | Claude Sonnet 4.6 | 1,633 | Feb 2026 | Model card | |
| 6 | Claude Opus 4.6 | 1,606 | Feb 2026 | Model card | |
| 7 | MiMo-V2.5-Pro | 1,581 | Apr 2026 | Model card | |
| 8 | DeepSeek-V4-Pro | 1,554 | Apr 2026 | Model card | |
| 9 | MiniMax-M2.7 | 1,495 | Mar 2026 | Model card | |
| 10 | Muse Spark | 1,444 | Apr 2026 | Model card | |
| 11 | MiMo-V2-Pro | 1,426 | Mar 2026 | Model card | |
| 12 | Step 3.7 Flash | 1,416 | May 2026 | Model card | |
| 13 | MiMo-V2-Omni | 1,410 | Mar 2026 | Model card | |
| 14 | GLM-5 | 1,409 | Feb 2026 | Model card | |
| 15 | DeepSeek-V4-Flash | 1,395 | Apr 2026 | Model card | |
| 16 | Gemini 3.1 Pro | 1,317 | Feb 2026 | Model card |
5 models have held number one since its release in Dec 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Anthropic
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Reported by the lab in its model card
Labs headline it less often now, last in Jun 2026
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| 9 Jun 2026 | Claude Fable 5 | 1932 | Knowledge work GDPval-AA |
| 29 May 2026 | Step 3.7 Flash | 1415.8 | GDPval-Stirrup 1415.8 (ii 45.8%) |
| 28 May 2026 | Claude Opus 4.8 | 1890 | Knowledge work GDPval-AA |
| 19 May 2026 | Gemini 3.5 Flash | 1656 | GDPval-AA (Elo) |
| 27 Apr 2026 | MiMo-V2.5-Pro | 1581 | GDPVal-AA (Elo) |
| 24 Apr 2026 | DeepSeek-V4-Pro | 1554 | GDPval-AA (Elo) |
| 20 Apr 2026 |  Qwen 3.6 Max (Preview) | 51 | GDPval-AA Real-World Valuable Task (51.0, scale not stated) |
| 8 Apr 2026 | Muse Spark | 1444 | GDPval-AA Elo (Office Tasks (Artificial Analysis)) |
| 18 Mar 2026 | MiMo-V2-Omni | 1410 | GDPVal (Elo) |
| 18 Mar 2026 | MiMo-V2-Pro | 1426 | GDPVal-AA |
| 18 Mar 2026 | MiniMax-M2.7 | 1495 | GDPval-AA (ELO score, text) |
| 19 Feb 2026 | Gemini 3.1 Pro | 1317 | GDPval-AA Elo |
| 17 Feb 2026 | Claude Sonnet 4.6 | 1633 | Office tasks GDPval-AA Elo |
| 5 Feb 2026 | Claude Opus 4.6 | 1606 | Office tasks GDPVal-AA Elo |
14 system cards report it, from 6 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 9 Jun 2026 | Claude Fable 5 system card | 1932 | GDPval-AA Elo as of June 6, 2026, no version named, run by Artificial Analysis, about 42 Elo ahead of Opus 4.8 (56% pairwise win rate) (Table 8.1.A, 8.17.7) | Capabilities |
| 28 May 2026 | Claude Opus 4.8 system card | 1890 | GDPval-AA Elo, run by Artificial Analysis (Table 8.1.A, 8.13.6) | Capabilities |
| 26 May 2026 | MiniMax-M2 technical report | 34 | GDPval-AA, MiniMax-M2 bar, printed label (Figure 9) | Capabilities |
| 26 May 2026 | MiniMax-M2.7 technical report | 50 | GDPval-AA, Artificial Analysis re-evaluation, score as printed (metric not stated) (Table 4) | Capabilities |
| 26 May 2026 | MiniMax-M2.5 technical report | 35 | GDPval-AA, Artificial Analysis re-evaluation, score as printed (metric not stated) (Table 4) | Capabilities |
| 19 May 2026 | Gemini 3.5 Flash model card | 1656 | GDPval-AA, Elo (results table, p.4) | Capabilities |
| 26 Apr 2026 | DeepSeek-V4-Pro technical report | 1554 | GDPval-AA (Elo), DS-V4-Pro Max (Table 6) | Capabilities |
| 26 Apr 2026 | DeepSeek-V4-Flash technical report | 1395 | GDPval-AA (Elo), DeepSeek-V4-Flash Max (Table 7) | Capabilities |
| 16 Apr 2026 | Claude Opus 4.7 system card | 1753 | GDPval-AA Elo, run by Artificial Analysis, printed on chart (8.10.5, Figure 8.10.5.A) | Capabilities |
| 19 Feb 2026 | Gemini 3.1 Pro model card | 1317 | GDPval-AA Elo (results table, p.4) | Capabilities |
| 17 Feb 2026 | GLM-5 technical report | 1409 | GDPval-AA Elo, recorded 15 Feb 2026 (Table 7) | Capabilities |
| 17 Feb 2026 | Claude Sonnet 4.6 system card | 1633 | GDPval-AA Elo, run by Artificial Analysis (Table 2.1.A, 2.8) | Capabilities |
| 5 Feb 2026 | Claude Opus 4.6 system card | 1606 | GDPval-AA Elo, run independently by Artificial Analysis, value printed on chart (Figure 2.10.A) | Capabilities |
| 30 Jan 2026 | Kimi K2.5 technical report | 41 | GDPVal-AA, Artificial Analysis leaderboard metric as of 2026-01-28 (Table 4, E.9) | Capabilities |
Versions
More in Economic work
- JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
- GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate.
Seed2.1 Pro 87.9% - AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Qwen3.8-Omni-Flash 75.3%
- OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%
- SpreadsheetBench In use Real spreadsheet manipulation questions from Excel forums. Output workbook is checked cell-by-cell against reference solutions. MiniMax-M3 89.3%