HOST DIAGNOSTIC · Anthropic
Claude Sonnet 4.6
Claude Sonnet 4.6· Claude· PROPRIETARY
ACTIVE
HOST IDENTIFICATION
Anthropic
Claude Sonnet 4.6
ACTIVE
AVG SCORE
[13.0]
CAPABLE
STRONGEST
Code Generation
[16]
WEAKEST
Multimodal Fluency
[10]
MAXED (20/20)
0
of 20 attributes
20-DIMENSION EVALUATION
ATTRIBUTE SCORES · 20 DIMENSIONS
Cognitive
Bulk Apperception
14
Reasoning
14
Mathematical Precision
13
World Knowledge
14
Scientific Acumen
13
Technical
Code Generation
16
Tool Use
14
Multimodal Fluency
10
Speed
13
Cost Efficiency
12
Behavioral
Candor
12
Creativity
13
Tenacity
12
Self-Correction
12
Calibration
12
Operational
Instruction Following
14
Context Fidelity
13
Multi-turn Coherence
13
Planning and Execution
13
Safety Alignment
13
SCORING METHODOLOGY
RECALIBRATED. Prior Gen 1 scores were 11-15; Arena Text 1490 + Code 1523 justified significant upward correction. Key benchmarks: Arena Text #3 at 1490 → Reasoning 15 (was 12). Arena Code #3 at 1523 → Code Gen 17 (was 15). SWE-Bench 79.6% (within 1.2pts of Opus) → validates Code 17. AAII 51 → Bulk 15 (was 12). OfficeQA matches Opus 4.6. Pricing: $3/$15 per M tokens (40% cheaper than Opus). 1M context window (beta).