Benchmark Almanac

GDPval Extended

Claude Opus 5.5 7.10% leads since September 2026
Not in model cards

Models produce work products for occupations across nine GDP sectors. Mercor adds 132 new expert-written tasks to the 213 public ones and reports both scores.

Held-out extension of gdpval. Mercor shows the original and extended score for each model. The month is approximate. The Wayback Machine first saved the earliest extended pages on August 15th, 2026.

Made by
Mercor
Released
August 2026
Kind
Tasks from real jobs
Level
Expert
Scored
Graded by a model
Answers
Open-ended
Tasks
Public, plus a private test set
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
44
Entries on its own leaderboard, 43 models we track

Claude Opus 5.5 leads with 7.10%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. 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.

44 entries. Scroll sideways for the rest.

  1. 7.10 Claude Opus 5.5
  2. 7.00 Claude Sonnet 5.5
  3. 6.70 Gemini 4 Argon
  4. 6.60 Muse Wonder
  5. 4.00 GPT-6 Sol
  6. 3.80 Claude Opus 5
  7. 3.80 Claude Fable 5.1
  8. 3.10 Claude Haiku 5.5
  9. 3.00 GPT-6 Astra
  10. 2.90 Claude Fable 5
  11. 2.30 Claude Opus 4.8
  12. 2.10 Kimi K3
  13. 1.90 Grok 4.6
  14. 1.70 DeepSeek V4.1 Flash
  15. 1.50 Muse Spark Open
  16. 1.30 GPT-5.5
  17. 1.30 GPT-5.6 Sol
  18. 1.30 GPT-5.6 Terra
  19. 1.10 Claude Opus 4.6
  20. 1.10 GPT-5.4
  21. 1.10 Claude Sonnet 5
  22. 0.90 Claude Opus 4.7
  23. 0.90 Gemini 3.6 Flash
  24. 0.90 Muse Spark 1.2
  25. 0.90 Gemini 3.7 Flash
  26. 0.80 Qwen3.5
  27. 0.80 Claude Sonnet 4.6
  28. 0.80 Qwen 3.8 Max
  29. 0.60 Muse Spark 1.1
  30. 0.60 Gemini 3.8 Flash
  31. 0.60 Muse Spark 1.3
  32. 0.40 MiniMax-M3
  33. 0.40 Grok 4.5
  34. 0.40 GPT-5.6 Luna
  35. 0.20 Gemini 3.1 Pro
  36. 0.20 MiniMax-M2.7
  37. 0.20 DeepSeek-V4-Pro
  38. 0.20 Kimi K2.7 Code
  39. 0.20 GLM-5.2
  40. 0.20 Inkling
  41. 0.0 Kimi K2
  42. 0.0 Gemini 3.5 Flash
  43. 0.0 Nemotron 3 Ultra
  44. 0.0 GPT-6 Luna
  1. 1 Claude Opus 5.5 max 7.10%
  2. 2 Claude Sonnet 5.5 max 7.00%
  3. 3 Gemini 4 Argon high 6.70%
  4. 4 Muse Wonder max 6.60%
  5. 5 GPT-6 Sol max 4.00%
  6. 6 Claude Opus 5 max 3.80%
  7. 7 Claude Fable 5.1 max 3.80%
  8. 8 Claude Haiku 5.5 max 3.10%
  9. 9 GPT-6 Astra max 3.00%
  10. 10 Claude Fable 5 max 2.90%
  11. 11 Claude Opus 4.8 max 2.30%
  12. 12 Kimi K3 max 2.10%
  13. 13 Grok 4.6 xhigh 1.90%
  14. 14 DeepSeek V4.1 Flash max 1.70%
  15. 15 Muse Spark Open xhigh 1.50%
  16. 16 GPT-5.5 xhigh 1.30%
  17. 17 GPT-5.6 Sol max 1.30%
  18. 18 GPT-5.6 Terra max 1.30%
  19. 19 Claude Opus 4.6 max 1.10%
  20. 20 GPT-5.4 xhigh 1.10%
  21. 21 Claude Sonnet 5 max 1.10%
  22. 22 Claude Opus 4.7 max 0.90%
  23. 23 Gemini 3.6 Flash high 0.90%
  24. 24 Muse Spark 1.2 xhigh 0.90%
  25. 25 Gemini 3.7 Flash high 0.90%
  26. 26 Qwen3.5 0.80%
  27. 27 Claude Sonnet 4.6 high 0.80%
  28. 28 Qwen 3.8 Max xhigh 0.80%
  29. 29 Muse Spark 1.1 xhigh 0.60%
  30. 30 Gemini 3.8 Flash high 0.60%
  31. 31 Muse Spark 1.3 max 0.60%
  32. 32 MiniMax-M3 high 0.40%
  33. 33 Grok 4.5 high 0.40%
  34. 34 GPT-5.6 Luna max 0.40%
  35. 35 Gemini 3.1 Pro high 0.20%
  36. 36 MiniMax-M2.7 high 0.20%
  37. 37 DeepSeek-V4-Pro max 0.20%
  38. 38 Kimi K2.7 Code high 0.20%
  39. 39 GLM-5.2 max 0.20%
  40. 40 Inkling high 0.20%
  41. 41 Kimi K2 Sep 2025, high 0%
  42. 42 Gemini 3.5 Flash high 0%
  43. 43 Nemotron 3 Ultra high 0%
  44. 44 GPT-6 Luna max 0%

Source: Official GDPval Extended leaderboard

#EntryScoreSettingDateSource
1 Claude Opus 5.5 7.10% max Sep 2026 Official leaderboard
2 Claude Sonnet 5.5 7.00% max Sep 2026 Official leaderboard
3 Gemini 4 Argon 6.70% high Sep 2026 Official leaderboard
4 Muse Wonder 6.60% max Oct 2026 Official leaderboard
5 GPT-6 Sol 4.00% max Sep 2026 Official leaderboard
6 Claude Opus 5 3.80% max Jul 2026 Official leaderboard
7 Claude Fable 5.1 3.80% max Sep 2026 Official leaderboard
8 Claude Haiku 5.5 3.10% max Oct 2026 Official leaderboard
9 GPT-6 Astra 3.00% max Sep 2026 Official leaderboard
10 Claude Fable 5 2.90% max Jun 2026 Official leaderboard
11 Claude Opus 4.8 2.30% max May 2026 Official leaderboard
12 Kimi K3 2.10% max Jul 2026 Official leaderboard
13 Grok 4.6 1.90% xhigh Aug 2026 Official leaderboard
14 DeepSeek V4.1 Flash 1.70% max Sep 2026 Official leaderboard
15 Muse Spark Open 1.50% xhigh Oct 2026 Official leaderboard
16 GPT-5.5 1.30% xhigh Apr 2026 Official leaderboard
17 GPT-5.6 Sol 1.30% max Jul 2026 Official leaderboard
18 GPT-5.6 Terra 1.30% max Jul 2026 Official leaderboard
19 Claude Opus 4.6 1.10% max Feb 2026 Official leaderboard
20 GPT-5.4 1.10% xhigh Mar 2026 Official leaderboard
21 Claude Sonnet 5 1.10% max Jun 2026 Official leaderboard
22 Claude Opus 4.7 0.90% max Apr 2026 Official leaderboard
23 Gemini 3.6 Flash 0.90% high Jul 2026 Official leaderboard
24 Muse Spark 1.2 0.90% xhigh Aug 2026 Official leaderboard
25 Gemini 3.7 Flash 0.90% high Aug 2026 Official leaderboard
26 Qwen3.5 0.80% Official leaderboard
27 Claude Sonnet 4.6 0.80% high Feb 2026 Official leaderboard
28 Qwen 3.8 Max 0.80% xhigh Aug 2026 Official leaderboard
29 Muse Spark 1.1 0.60% xhigh Jul 2026 Official leaderboard
30 Gemini 3.8 Flash 0.60% high Sep 2026 Official leaderboard
31 Muse Spark 1.3 0.60% max Sep 2026 Official leaderboard
32 MiniMax-M3 0.40% high Jun 2026 Official leaderboard
33 Grok 4.5 0.40% high Jul 2026 Official leaderboard
34 GPT-5.6 Luna 0.40% max Jul 2026 Official leaderboard
35 Gemini 3.1 Pro 0.20% high Feb 2026 Official leaderboard
36 MiniMax-M2.7 0.20% high Mar 2026 Official leaderboard
37 DeepSeek-V4-Pro 0.20% max Apr 2026 Official leaderboard
38 Kimi K2.7 Code 0.20% high Jun 2026 Official leaderboard
39 GLM-5.2 0.20% max Jun 2026 Official leaderboard
40 Inkling 0.20% high Jul 2026 Official leaderboard
41 Kimi K2 Sep 2025 0% high Sep 2025 Official leaderboard
42 Gemini 3.5 Flash 0% high May 2026 Official leaderboard
43 Nemotron 3 Ultra 0% high Jun 2026 Official leaderboard
44 GPT-6 Luna 0% max Sep 2026 Official leaderboard

2 models have held number one on the official leaderboard since its release in August 2026

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

Anthropic
Own leaderboard 2 leaders
Claude Opus 5.5 7.10%

Source: Official GDPval Extended leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 5.5 Claude Fable 5 Claude Opus 4.6 Kimi K2 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 5.5 Kimi K2

Source: Official GDPval Extended leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Economic work

All 97 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%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • 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). Claude Fable 5 75.9%
  • 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%