JOHNCLICK.AI
HOST DIAGNOSTIC · Anthropic

Claude Opus 4.7

Claude Opus 4.7 · ACTIVE· Claude· CLOSED
ACTIVE
ATTRIBUTE MATRIX · ATRBT GROUP 01
Bulk Apperception
[14]
Reasoning
[15]
Math Precision
[15]
Code Generation
[18]
World Knowledge
[15]
Scientific Acumen
[16]
Instruction Following
[16]
Context Fidelity
[13]
Candor
[13]
Creativity
[14]
Multi-turn Coherence
[14]
Multimodal Fluency
[15]
Tool Use
[15]
Planning
[15]
Self-Correction
[15]
Tenacity
[15]
Safety Alignment
[16]
Calibration
[13]
Speed
[7]
Cost Efficiency
[3]
HOST TELEMETRY
AVG
13.8
/20
BEST
18
Code Generation
WEAK
3
Cost Efficiency
PARAMETERS
Undisclosed
CONTEXT
—
PRICING
—
STATUS
ACTIVE
ATTRIBUTE SCORES · 20 DIMENSIONS
Cognitive
Bulk Apperce…
14
Reasoning
15
Mathematical…
15
World Knowle…
15
Scientific A…
16
Technical
Code Generat…
18
Tool Use
15
Multimodal F…
15
Speed
7
Cost Efficie…
3
Behavioral
Candor
13
Creativity
14
Tenacity
15
Self-Correct…
15
Calibration
13
Operational
Instruction …
16
Context Fide…
13
Multi-turn C…
14
Planning and…
15
Safety Align…
16
SCORING METHODOLOGY
CALIBRATED per ADR-NM108 (April 22, 2026). Benchmark-derived scores adjusted with 100% personal evaluation weight. Key benchmarks: SWE-bench Verified 87.6%. GPQA Diamond 94.2%. Arena Code #1 at 1560 ELO. Arena Text 1497 ELO. MMLU-Pro 89.87%. ARC-AGI-2 75.83%. MCP-Atlas 77.3%. REGRESSIONS: BrowseComp 83 to 79. Terminal-Bench trails GPT-5.4. New tokenizer +35% effective cost. Combative literalism. Sources: Anthropic, BenchLM, Vellum, Nate B Jones, Arena.ai, Epoch AI, Vals AI.
← ALL HOSTS
COMPARE IN MATRIX →KNOWLEDGE GRAPH →