DAYJOB: Law Can AI do a lawyer’s job?

Professional Graphical Reasoning

Chartography

Chartography is our benchmark for professional chart understanding. It tests whether frontier models can read the Kaplan-Meier curves, candlestick charts, contour maps, Sankey diagrams, Bode plots, and other specialized graphics that professionals use to make real decisions every day. It evaluates visual perception, domain-aware interpretation, and multi-step graphical reasoning.

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.
Chartography Leaderboard
1
GPT
6.2 Astra (Max reasoning)
90.7
%
1
Claude
Opus 5.5 (Adaptive/Max)
89.6
%
1
GPT
6 Astra (Max reasoning)
88.6
%
1
GPT
6.2 Sol (Max reasoning)
87.7
%
1
Gemini
4 Hydrogen (xHigh reasoning)
87.2
%
1
Gemini
4 Neon (High reasoning)
86.6
%
1
Gemini
4 Carbon (High reasoning)
85.5
%
1
Muse
Spark 1.3 (Max reasoning)
84.6
%
1
Claude
Opus 5 (Adaptive/Max)
84.4
%
1
Gemini
3.8 Flash (High reasoning)
81.7
%
1
Claude
Fable 5.2 (Adaptive/Max)
80.8
%
1
Claude
Fable 5.1 (Adaptive/Max)
79.6
%
1
GPT
6 Sol (Max reasoning)
79.4
%
1
Muse
Ember 1.0 (Max reasoning)
79.3
%
1
Claude
Fable 5 (Adaptive/Max)
79
%
1
GPT
5.6 Sol (Max reasoning)
78.6
%
1
DeepSeek
V4 Flash (Max reasoning)
77.1
%
1
Inkling
2 Small (xHigh reasoning)
75.6
%
1
Qwen
4 Max (xHigh reasoning)
75.3
%
1
Qwen
3.8 Max (xHigh reasoning)
75.2
%
1
Gemini
3.7 Flash (High reasoning)
74.9
%
1
Grok
4.7 (xHigh reasoning)
74.6
%
1
GLM
5.3 Flash (Max reasoning)
72.7
%
1
GPT
5.6 Luna (Max reasoning)
72.2
%
1
GPT
6 Luna (Max reasoning)
68
%
1
Muse
Spark 1.2 (xHigh reasoning)
66.7
%
1
Kimi
K3 (Max reasoning)
66.4
%
1
GLM
5.3 (Max reasoning)
64.2
%
1
Claude
Sonnet 5 (Adaptive/Max)
64
%
1
Hy
Hy4 Preview (High reasoning)
59.4
%
1
GPT
5.6 Terra (xHigh reasoning)
58.8
%
1
Gemini
3.1 Pro (High reasoning)
54.2
%
1
Hy
Hy4 (High reasoning)
49.2
%
1
Grok
4.6 (xHigh reasoning)
44.4
%
1
MiniMax
M3.1 Flash (Max reasoning)
39.8
%
1
Kimi
K2.7 Code (Max reasoning)
31.5
%
1
DeepSeek
V4 Pro (Max reasoning)
21.1
%
1
Hy
Hy3 (High reasoning)
20.3
%
1
Muse
Glimmer 30B (xHigh reasoning)
15.4
%
1
Inkling
Inkling (High reasoning)
15
%
1
Mistral
Large 3
7.3
%
1
Nemotron
3 Ultra
6.8
%

Score vs. cost per trial

Score vs. generated tokens

Stay Posted