BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases4

GDPx leaderboard

As of 2 Nov 2026, the top score on GDPx is 12.0% by Claude Opus 5.5 (Anthropic), ahead of Gemini 4 Carbon at 10.9%, across 20 model configurations with a published result (GDPx by SemiAnalysis, result dated 1 Nov 2026). SemiAnalysis' agentic benchmark: 206 long tasks across 11 fields such as chip design, cybersecurity, biology and AI research, each run in the model's own coding harness. The score is the share of tasks passed (three attempts, every field weighted equally). GDPx is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 2 Nov 2026)

  • Claude Opus 5.5 holds the top GDPx score at 12.0% (GDPx by SemiAnalysis, 1 Nov 2026).
  • Gemini 4 Carbon is second on GDPx at 10.9% (GDPx by SemiAnalysis, 1 Nov 2026).
  • GDPx: SemiAnalysis' agentic benchmark: 206 long tasks across 11 fields such as chip design, cybersecurity, biology and AI research, each run in the model's own coding harness. The score is the share of tasks passed (three attempts, every field weighted equally) (GDPx by SemiAnalysis).
  • 20 model configurations have a published GDPx result on BenchLeader (BenchLeader, 2 Nov 2026).
  • GDPx does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

SemiAnalysis' agentic benchmark: 206 long tasks across 11 fields such as chip design, cybersecurity, biology and AI research, each run in the model's own coding harness. The score is the share of tasks passed (three attempts, every field weighted equally).

Published by
GDPx by SemiAnalysis
Category
Agents & tools
Index weight
Reference only
Models
20
Data as of
2 Nov 2026

© SemiAnalysis. GDPx (https://gdpx.semianalysis.com). Headline scores read daily from the public leaderboard and shown with credit and a link; not endorsed by SemiAnalysis.

Top 15
  1. 1Claude Opus 5.5 (max)12.0%
  2. 2Gemini 4 Carbon (high)10.9%
  3. 3GPT-6 Astra (max)9.6%
  4. 4Muse Wonder Full (max)9.0%
  5. 5Claude Fable 5.1 (max)8.5%
  6. 6GPT-6.1 Terra (max)8.5%
  7. 7GPT-6.1 Sol (max)8.4%
  8. 8Claude Sonnet 5.5 (max)8.2%
  9. 9Gemini 4 Argon (high)7.3%
  10. 10Muse Wonder (max)6.8%
  11. 11DeepSeek V4.1 Flash 1017 (max)5.4%
  12. 12GPT-6 Luna (max)5.1%
  13. 13Claude Haiku 5.5 (max)4.9%
  14. 14MiniMax M3.1 Flash (xhigh)4.0%
  15. 15Grok 4.7 (xhigh)3.9%
20 of 20
#
1Claude Opus 5.5max12.0%69.22026-11-01
2NewGemini 4 Carbonhigh10.9%68.92026-11-01
3GPT-6 Astramax9.6%67.72026-11-01
4Muse Wonder Fullmax9.0%62.92026-11-01
5Claude Fable 5.1max8.5%67.12026-11-01
6NewGPT-6.1 Terramax8.5%65.22026-11-01
7GPT-6.1 Solmax8.4%67.22026-11-01
8Claude Sonnet 5.5max8.2%65.12026-11-01
9Gemini 4 Argonhigh7.3%65.92026-11-01
10Muse Wondermax6.8%62.22026-11-01
11DeepSeek V4.1 Flash 1017max5.4%59.22026-11-01
12GPT-6 Lunamax5.1%56.72026-11-01
13Claude Haiku 5.5max4.9%56.92026-11-01
14MiniMax M3.1 Flashxhigh4.0%–2026-11-01
15Grok 4.7xhigh3.9%58.22026-11-01
16Kimi K3max3.7%62.12026-11-01
17Qwen3.8 Maxmax3.5%58.82026-11-01
18GLM 5.3max2.7%59.02026-11-01
19Inklingmax2.3%–2026-11-01
20Muse Spark Openmax1.7%–2026-11-01

More agents & tools benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on GDPx

Cite as: BenchLeader, “GDPx leaderboard”, data as of 2 Nov 2026, https://www.benchleader.com/benchmarks/gdpx.

GDPx: questions

What does GDPx measure?
SemiAnalysis' agentic benchmark: 206 long tasks across 11 fields such as chip design, cybersecurity, biology and AI research, each run in the model's own coding harness. The score is the share of tasks passed (three attempts, every field weighted equally). Scores are reported in percent of tasks solved; higher is better.
What is the top score on GDPx?
As of 2 Nov 2026, the highest score on GDPx is 12.0% by Claude Opus 5.5 (result dated 1 Nov 2026), ahead of Gemini 4 Carbon at 10.9%. Scores come from GDPx by SemiAnalysis and are re-read daily.
Which AI model leads GDPx?
Claude Opus 5.5 leads GDPx with 12.0% as of 2 Nov 2026 (result dated 1 Nov 2026), ahead of Gemini 4 Carbon at 10.9%.
How many models have GDPx results?
20 model configurations have a GDPx result on BenchLeader, all taken from GDPx by SemiAnalysis.
Who runs GDPx and how often is it updated?
GDPx is published by GDPx by SemiAnalysis. BenchLeader re-reads the published results once a day and records the date each result was published.
Does GDPx count toward the BenchLeader Index?
No. GDPx is shown for reference but left out of the composite index.