RL Environments and the Hierarchy of Agentic Capabilities
Our RL environment run on 9 models revealed the core capabilities all agents need to master: tool use, planning, adaptability, groundedness, and common sense.
GDP.pdf Leaderboard
1
GPT
6.1 Sol (Max reasoning)
35
%
1
GPT
6 Astra (Max reasoning)
34.2
%
1
Muse
Wonder Full (Max reasoning)
33.2
%
1
Claude
Sonnet 5.5 (Adaptive/Max)
31.4
%
1
Muse
Wonder (Max reasoning)
30.8
%
1
GPT
5.6 Sol (Max reasoning)
30.7
%
1
Claude
Opus 5.5 (Adaptive/Max)
30.6
%
1
Claude
Fable 5 (Adaptive/Max)
29.8
%
1
Gemini
4 Argon (High reasoning)
28.4
%
1
Claude
Fable 5.1 (Adaptive/Max)
27.6
%
1
Muse
Spark 1.3 (xHigh reasoning)
27.6
%
1
GPT
6 Sol (Max reasoning)
26.4
%
1
GPT
5.5 (xHigh reasoning)
26
%
1
Claude
Opus 4.8 (Adaptive/Max)
24
%
1
Claude
Opus 5 (Adaptive/Max)
24
%
1
Gemini
3.7 Flash (High reasoning)
23.8
%
1
Gemini
3.8 Flash (High reasoning)
23.2
%
1
Qwen
3.8 Max (xHigh reasoning)
23.2
%
1
GPT
6 Luna (Max reasoning)
23
%
1
Grok
4.7 (xHigh reasoning)
22.8
%
1
Claude
Opus 4.7 (Adaptive/Max)
21
%
1
DeepSeek
V4.1 Flash (Max reasoning)
19.8
%
1
Kimi
K3 (Max reasoning)
19
%
1
Claude
Sonnet 4.6 (Adaptive/Max)
18
%
1
Grok
4.6 (xHigh reasoning)
17.2
%
1
Gemini
3.1 Pro (High reasoning)
17
%
1
Qwen
3.8 Flash (xHigh reasoning)
16.6
%
1
DeepSeek
V4 Flash Vision (experimental) (Max reasoning)
16.2
%
1
Muse
Spark 1.2 (xHigh reasoning)
16
%
1
Gemini
3.6 Flash (High reasoning)
14
%
1
GLM
5.3 Flash (Max reasoning)
14
%
1
Grok
4.5 (High reasoning)
14
%
1
Kimi
K2.6 (Thinking on)
12
%
1
Muse
Glimmer 30B (xHigh reasoning)
11.8
%
1
Gemini
3.5 Flash-Lite (High reasoning)
10
%
1
Gemini
3 Flash (High reasoning)
10
%
1
Grok
4.3 (High reasoning)
8
%
1
Mistral
Large 3
2
%
1
Nemotron
3 Nano Omni
2
%
VIEW ALL BENCHMARKS
Agentic Coding
sudo L7
Coding agents are getting very good at being L3s. But do they have the judgment to become L7s? Across 60 tasks, agents must navigate unfamiliar systems, choose architectures, spot hidden risks, and know when technically correct code is still the wrong thing to ship.
Rank
Model
Score
41.7
%
Claude Sonnet 5.5 (Adaptive/Max)
Anthropic
41.7
%
Claude Opus 5.5 (Adaptive/Max)
Anthropic
33.9
%
GPT 6 Astra (Max)
OpenAI
31.1
%
Claude Fable 5.1 (Adaptive/Max)
Anthropic
28.9
%
GPT 6.1 Sol (Max)
OpenAI
20.6
%
Qwen 3.8 Max
Alibaba Cloud
Agentic Finance
DAYJOB: Finance
DAYJOB: Finance measures long-horizon finance agents across corporate finance, banking, credit, investing, and real assets. It's built around the messy reality of knowledge work: realistic requests, complex environments, and end-to-end professional judgment.
Rank
Model
Score
23.9
%
Claude Opus 5.5 (Adaptive/Max)
Anthropic
21.5
%
GPT 6 Astra (Max)
OpenAI
20.3
%
Gemini 4 Argon (High)
Google
19.8
%
Claude Fable 5.1 (Adaptive/Max)
Anthropic
14.8
%
Muse Spark 1.3 (Max)
Meta
14.5
%
Grok 4.7 (xHigh)
xAI