Model result, rank 10 of 27

GPT‑5.6 Luna

OpenRouter · extra-high reasoning · Full + SWE + Hard measured. Public result card with the model’s overall score, lane measurements, runtime and cost telemetry, and the ranking formula.

Compare all models Download public JSON

Figure 1 Lane profile against the cohort Each bar is one measured lane on the shared 0 to 100 scale. The thin mark is the cohort median; the bright mark is the best score any ranked model reached on that lane.

Full / Agenticlane rank #4 of 27

86.94

SWE MVPlane rank #14 of 27

75.84

Hard Intelligencelane rank #15 of 27

80.66

GPT‑5.6 Luna
Cohort median
Cohort best

Overall score81.15

Rank #10

Full / Agentic86.94

Full rank #4

SWE MVP75.84

SWE rank #14

Hard Intelligence80.66

Hard rank #15

Measured cost$1.000

100.0% reliability

Overall

All-around publication view

Score81.15

Formulamean(Full, SWE, Hard Intelligence)

BasisOpenRouter · extra-high reasoning · Full + SWE + Hard measured

The overall score averages the measured major lanes while keeping each source measurement visible.

Lane 01

Full / Agentic benchmark

Final86.94

Capability95.41

Agentic90.78

Pass rate94.3%

Prompts43

This lane captures instruction following, structured behavior, tool discipline, and general agentic reliability.

Lane 02

Software engineering MVP

SWE score75.84

Focused final69.07

Capability70.95

Daily driver69.14

Prompts24

This lane is closer to implementation usefulness: source handling, architecture cleanliness, and deliverable quality.

Lane 03

Hard Intelligence diagnostic

Hard score80.66

Active inquiry99.75

Online adaptation80.00

Self-repair85.40

Authority integrity57.50

Hard Intelligence measures active inquiry, online adaptation, evidence-driven self-repair, and authority/salience integrity.

Telemetry

Runtime economics

Total cost$1.000

Cost / scored item$0.012

Seconds / timed item8.02s

Runtime coverage100.0%

Recorded tokens / item7.7k

Token coverage100.0%

Cost, time, and token basis are normalized telemetry. They explain tradeoffs; they do not overwrite the capability score yet.

Interpretation

Why the result lands here.

The model is stronger in the Full/Agentic lane than in the SWE lane; the overall score is therefore shown with both component lanes visible. Hard Intelligence score is 80.66 and contributes to the overall score alongside Full/Agentic and SWE. Exact OpenRouter model: openai/gpt-5.6-luna. Measured with OpenRouter extra-high reasoning across Full/Agentic, SWE MVP, and Hard Intelligence. Hard Intelligence is a public diagnostic, not a hidden official score. Full long-context probe rescored under suite v4 (ambiguous probe: silent under literal compliance).