JOHNCLICK.AI
HOST DIAGNOSTIC · Google

Gemini 2.5 Deep Think

Gemini 2.5 Deep Think · ACTIVE· Gemini· CLOSED
ACTIVE
ATTRIBUTE MATRIX · ATRBT GROUP 01
Bulk Apperception
[16]
Reasoning
[17]
Math Precision
[17]
Code Generation
[15]
World Knowledge
[15]
Scientific Acumen
[17]
Instruction Following
[15]
Context Fidelity
[16]
Candor
[15]
Creativity
[14]
Multi-turn Coherence
[15]
Multimodal Fluency
[14]
Tool Use
[13]
Planning
[16]
Self-Correction
[14]
Tenacity
[15]
Safety Alignment
[14]
Calibration
[15]
Speed
[6]
Cost Efficiency
[9]
HOST TELEMETRY
AVG
14.4
/20
BEST
17
Reasoning
WEAK
6
Speed
PARAMETERS
Undisclosed
CONTEXT
—
PRICING
—
STATUS
ACTIVE
ATTRIBUTE SCORES · 20 DIMENSIONS
Cognitive
Bulk Apperce…
16
Reasoning
17
Mathematical…
17
World Knowle…
15
Scientific A…
17
Technical
Code Generat…
15
Tool Use
13
Multimodal F…
14
Speed
6
Cost Efficie…
9
Behavioral
Candor
15
Creativity
14
Tenacity
15
Self-Correct…
14
Calibration
15
Operational
Instruction …
15
Context Fide…
16
Multi-turn C…
15
Planning and…
16
Safety Align…
14
SCORING METHODOLOGY
CALIBRATED per ADR-NM108 (April 2026 snapshot). Deep reasoning variant of Gemini 2.5 family. ARC-AGI-2: 45.1% (Gemini 3 Deepthink variant). Math + Science scored high (17) as extended reasoning model. Speed scored low (implicit slow reasoning). Cost 9 (extended thinking burns tokens). On LM Council benchmark suite.
← ALL HOSTS
COMPARE IN MATRIX →KNOWLEDGE GRAPH →