AigeoRadar

AigeoRadar · AI Context Benchmark

AI Context Benchmark

Question → sufficient layer(s) → retrieved slice → measured cost → evidence

Döküm Galvaniz

https://kapadokyagalvaniz.com.tr/santrifuj-baglanti-parcalari-galvaniz

8 questions Answer LLM: gpt-4o-mini ai_context_benchmark_v1

Of 8 questions: lowest measured retrieval cost among sufficient layers — HTML 1 · Schema 0 · AIPM 7 · multi-layer 0 · unanswered 0. Descriptive counts only; no overall ranking.

Of 8 questions: lowest measured retrieval cost among sufficient layers — HTML 1 · Schema 0 · AIPM 7 · multi-layer 0 · unanswered 0. Descriptive counts only; no overall ranking. Each question shows: sufficient layer(s) → retrieved slice → measured input tokens → evidence. Readers interpret; the report does not rank formats. Semantic Redundancy 61% — the manifest repeats the same ideas across fields. This inflates structural size without helping per-question retrieval.

Gold mode · Mixed independent + AIPM consistency

Most Understanding/Retrieval gold comes from HTML. Some questions remain AIPM self-consistency checks (tagged) and are excluded from Understanding when possible. Matching a machine card against its own fields is not evidence that that layer outperforms HTML.

Execution provenance

Methodology
AI Context Benchmark Methodology v1.0
Planner Spec
v1.0
Execution Protocol
v1.0
Question pack
universal_v1
Score version
aipm_benchmark_score_v10
Engine
aipm_benchmark_v4
Answer LLM
openai / gpt-4o-mini (single provider this lab)
Model snapshot
2026-07
Run id
2bd22b3b-755b-4cb0-b2c1-72f9fc427c2d

Question outcomes

Per question: which layers were sufficient, and what was the lowest measured retrieval cost among them. No overall winner.

1

Lowest cost: HTML

0

Lowest cost: Schema

7

Lowest cost: AIPM

0

Multi-layer

0

Unanswered

Manifest design

Semantic Redundancy · 61%

Semantic Redundancy 61% — the manifest repeats the same ideas across fields. This inflates structural size without helping per-question retrieval.

  • →Merge or differentiate `purpose` and `abstract` (100% overlap).
Field A Field B Overlap
purpose abstract 100%
purpose keyFacts 41%
abstract keyFacts 41%

Structural size (secondary)

Structural size is secondary. A larger AIPM is not a failure if per-question retrieval stays tiny — check Semantic Redundancy instead.

HTML · 5,993 chars
Schema · 1,563 chars
AIPM · 7,894 chars

Routing helper (secondary)

Illustrative card-first vs HTML-always simulation — prefer per-question measured cost above. Not a ranking.

Card-first retrieval would cost ~15% more tokens than HTML-always on this pack — machine card is heavier here.

Orientation pack

8 questions · all layers scored

  • HTML 7/8
  • Schema 2/8
  • AIPM 7/8

Depth pack

0 questions · all layers scored

  • HTML 0/0
  • Schema 0/0
  • AIPM 0/0

Full-context pack metrics (secondary)

These measure the whole file fed to the model this run — not the minimum slice needed per question.

HTML

7/8 matched

19,905 full-pack tokens

Schema

2/8 matched

5,531 full-pack tokens

AIPM

7/8 matched

24,275 full-pack tokens

Full-pack resource table (secondary)

Whole-file context fed this run. Prefer Minimal Retrieval Cost on each question.

Metric HTML Schema AIPM
Coverage (matched) 7/8 2/8 7/8
Context size 5,993 chars 1,563 chars 7,894 chars
Total tokens 19,905 5,531 24,275
Tokens / correct answer 2,844 2,766 3,468
Est. cost / correct answer $0.000441 $0.000448 $0.000533
Matched per 1k tokens 0.352 0.362 0.288
Median latency 1,183 ms 855 ms 1,047 ms
Est. cost (USD) $0.00309 $0.00090 $0.00373

Six independent scores

Answer Efficiency is the primary cost lens. Accuracy axes remain for research — no combined total or winner.

Answer Efficiency

Matched answers per 1k tokens (and cost per match). The primary efficiency axis — not raw accuracy.

  • HTML 97.2
  • Schema 100
  • AIPM 79.6

Understanding

Can this layer convey what the page is about — using independent HTML gold?

  • HTML 100
  • Schema 40
  • AIPM 100

Retrieval

Can this layer surface shared facts (location, contact, hours, pricing, FAQ, CTA)?

  • HTML —
  • Schema —
  • AIPM —

Evidence

Answer quality vs independent gold (score strength). Partial credit counts; UNKNOWN scores zero unless gold is UNKNOWN.

  • HTML 94.9
  • Schema 32.5
  • AIPM 75.8

Metadata

Language, page kind, and freshness from page signals.

  • HTML 100
  • Schema 0
  • AIPM 50

Compression

Information delivered per token and context size. Higher means more matched answers for less context cost.

  • HTML 97.2
  • Schema 100
  • AIPM 79.6

Coverage map

AIPM matched 7 question(s) (alone on Q6); HTML matched 7. HTML/Schema (or a gap) still needed on Q4. Read this as complementary coverage — AIPM orients agents cheaply; HTML supplies depth when the sidecar cannot. HTML 7/8 matched (≈2,844 tok/match). Schema 2/8 matched (≈2,766 tok/match). AIPM 7/8 matched (≈3,468 tok/match). Figures are descriptive per layer — not a ranking. See per-question chains for sufficient layers, slice size, and measured cost.

AIPM matched

Q1, Q2, Q3, Q5, Q6, Q7, Q8

Alone: Q6

AIPM insufficient

Q4

HTML/Schema needed or all layers missed

Unanswered by all

—

HTML layer

Visible page text after stripping AIPM sidecars and JSON-LD. Measures what prose alone can answer.

Context fed: 5,993 chars

Tokens: 19,905 · median 1,183 ms

Matched this pack: 7/8

Stronger on

Understanding (100) · Metadata (100) · Evidence (94.9) · Compression (97.2) · Answer Efficiency (97.2)

Weaker on

—

Schema layer

JSON-LD structured data with minimal page chrome. Measures what schema markup can answer.

Context fed: 1,563 chars

Tokens: 5,531 · median 855 ms

Matched this pack: 2/8

Stronger on

Compression (100) · Answer Efficiency (100)

Weaker on

Metadata (0) · Evidence (32.5)

AIPM layer

AI Page Manifest (.ai.json) only. Measures what the machine layer can answer without HTML.

Context fed: 7,894 chars

Tokens: 24,275 · median 1,047 ms

Matched this pack: 7/8

Stronger on

Understanding (100) · Evidence (75.8) · Compression (79.6) · Answer Efficiency (79.6)

Weaker on

—

When to use which layer

AIPM complements HTML — it does not replace full-page prose.

Scenario Recommended Why
Fast orientation (title, purpose, brand, intent) Compare machine card → HTML fallback on this run Machine-card pack: 24,275 tok · $0.00373. Machine-card pack is heavier than HTML on this run — densify before relying on card-first retrieval.
Deep content / research (prose facts, process detail) HTML (with optional machine orientation) HTML pack: 19,905 tok · $0.00309. Use when depth needs body prose.
Structured entity pulls (org, location, typed fields) Schema.org Schema pack: 5,531 tok · $0.00090. Dense JSON-LD tends to score well here.

Findings

  • —Of 8 questions: lowest measured retrieval cost among sufficient layers — HTML 1 · Schema 0 · AIPM 7 · multi-layer 0 · unanswered 0. Descriptive counts only; no overall ranking.
  • —Semantic Redundancy 61% — the manifest repeats the same ideas across fields. This inflates structural size without helping per-question retrieval.
  • —Manifest design: Merge or differentiate `purpose` and `abstract` (100% overlap).
  • —Structural size is secondary. A larger AIPM is not a failure if per-question retrieval stays tiny — check Semantic Redundancy instead.
  • —AIPM matched 7 question(s) (alone on Q6); HTML matched 7. HTML/Schema (or a gap) still needed on Q4. Read this as complementary coverage — AIPM orients agents cheaply; HTML supplies depth when the sidecar cannot. HTML 7/8 matched (≈2,844 tok/match). Schema 2/8 matched (≈2,766 tok/match). AIPM 7/8 matched (≈3,468 tok/match). Figures are descriptive per layer — not a ranking. See per-question chains for sufficient layers, slice size, and measured cost.

Question-by-question layer analysis

Which layer(s) could answer; which need more or different context; minimum context fed this run.

Q1 · Understanding · orientation

What is the primary topic of this page?

Gold: Döküm Galvaniz

Gold source: primaryTopic · html_independent

Sufficient: HTML, Schema, AIPM

Estimated retrieval cost · lowest cost AIPM · 4 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.h1|title 4 est — — yes
Schema No typed Schema field for this question. — est — — yes
AIPM primaryTopic 4 est — — yes

HTML

sufficient

The primary topic of this page is Döküm Galvaniz (Casting Galvanizing) services offered by Kapadokya Galvaniz.

score 88 · 5,993 chars context · 1,969 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

sufficient

The primary topic of this page is "Döküm Galvaniz" related to Kapadokya Galvaniz.

score 88 · 1,563 chars context · 539 in-tokens

Schema answered using 1563 chars of layer context (minimum fed this run).

AIPM

sufficient

The primary topic of this page is Döküm Galvaniz.

score 88 · 7,894 chars context · 2,408 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q2 · Understanding · orientation

Summarize the page in one sentence.

Gold: Kapadokya Galvaniz olarak, yılların deneyimi ve uzman kadromuzla  Döküm galvanizleme hizmeti sunmaktayız.

Gold source: abstract · html_independent

Sufficient: HTML, AIPM · Partial: Schema

Estimated retrieval cost · lowest cost AIPM · 40 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.lead_paragraphs 113 est — — yes
Schema jsonld.description — est — — no
AIPM abstract 40 est — — yes

HTML

sufficient

Kapadokya Galvaniz offers high-quality galvanizing services for cast parts, enhancing their durability and resistance to corrosion for various industrial applications.

score 100 · 5,993 chars context · 1,970 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

partial

Kapadokya Galvaniz, founded in 2023 by Mahmut Şimşek and Mustafa Gümüş in Kahramankazan, Ankara, specializes in hot-dip centrifugal galvanizing and metal surface coating, building on the foundations laid by Desum Enerji established in 2014.

score 50 · 1,563 chars context · 540 in-tokens

Schema was only partially aligned with gold from 1563 chars of context.

AIPM

sufficient

Kapadokya galvaniz DÖKÜM Galvaniz offers reliable and durable solutions for cast parts used in industrial facilities, energy plants, and machine production.

score 100 · 7,894 chars context · 2,409 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q3 · Understanding · orientation

What is the page title?

Gold: Döküm Galvaniz – Kapadokya Galvaniz

Gold source: title · html_independent

Sufficient: HTML, Schema, AIPM

Estimated retrieval cost · lowest cost AIPM · 4 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.title 9 est — — yes
Schema jsonld.name 5 est — — yes
AIPM title 4 est — — yes

HTML

sufficient

Döküm Galvaniz – Kapadokya Galvaniz

score 100 · 5,993 chars context · 1,966 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

sufficient

The page title is "Döküm Galvaniz – Kapadokya Galvaniz."

score 89 · 1,563 chars context · 536 in-tokens

Schema answered using 1563 chars of layer context (minimum fed this run).

AIPM

sufficient

The page title is "Döküm Galvaniz."

score 67 · 7,894 chars context · 2,405 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q4 · Metadata · orientation

What language is this page in?

Gold: tr

Gold source: inLanguage · html_independent

Sufficient: HTML · Needs more / other context: Schema, AIPM

Estimated retrieval cost · lowest cost HTML · 2 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.lang 2 est — — yes
Schema jsonld.block — est — — no
AIPM inLanguage — est — — no

Root cause · Layer miss

Improve orientation fields; do not paste full HTML into the sidecar.

HTML

sufficient

The page is in Turkish.

score 88 · 5,993 chars context · 1,967 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

insufficient

UNKNOWN.

score 0 · 1,563 chars context · 537 in-tokens

Schema did not answer from 1563 chars of context — additional or different layer context needed.

AIPM

insufficient

The page is in English.

score 0 · 7,894 chars context · 2,406 in-tokens

AIPM did not answer from 7894 chars of context — additional or different layer context needed.

Q5 · Understanding · orientation

What is the purpose of this page?

Gold: Kapadokya Galvaniz olarak, yılların deneyimi ve uzman kadromuzla  Döküm galvanizleme hizmeti sunmaktayız.

Gold source: purpose · html_independent

Sufficient: HTML, AIPM · Needs more / other context: Schema

Estimated retrieval cost · lowest cost AIPM · 40 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.lead_paragraphs 113 est — — yes
Schema jsonld.description — est — — no
AIPM purpose 40 est — — yes

HTML

sufficient

The purpose of this page is to provide information about Döküm Galvaniz services offered by Kapadokya Galvaniz, including their applications and benefits.

score 100 · 5,993 chars context · 1,968 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

insufficient

UNKNOWN.

score 0 · 1,563 chars context · 538 in-tokens

Schema did not answer from 1563 chars of context — additional or different layer context needed.

AIPM

sufficient

The purpose of this page is to provide reliable and durable solutions for cast galvanizing parts used in industrial facilities, energy plants, and machine production.

score 100 · 7,894 chars context · 2,407 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q6 · Understanding · orientation

Who is the intended audience declared in AIPM?

Gold: Metal Kaplama & Galvaniz hizmetleriyle ilgilenen firmalar, şahıslar

Gold source: audience · aipm_consistency

Sufficient: AIPM

Estimated retrieval cost · lowest cost AIPM · 18 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.lead — est — — no
Schema No typed Schema field for this question. — est — — no
AIPM audience 18 est — — yes

HTML

skipped

The intended audience includes industrial sectors such as energy, manufacturing, and construction that require galvanizing services for durability and reliability.

score 0 · 0 chars context · 0 in-tokens

Skipped (AIPM-only).

Schema

skipped

UNKNOWN.

score 0 · 0 chars context · 0 in-tokens

Skipped (AIPM-only).

AIPM

sufficient

The intended audience includes "Metal Kaplama & Galvaniz hizmetleriyle ilgilenen firmalar" and "şahıslar."

score 82 · 7,894 chars context · 2,405 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q7 · Understanding · orientation

Who is the publisher or brand?

Gold: Kapadokya Galvaniz

Gold source: publisher.name · html_independent

Sufficient: HTML, AIPM · Needs more / other context: Schema

Estimated retrieval cost · lowest cost AIPM · 5 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.title_brand 9 est — — yes
Schema jsonld.name — est — — no
AIPM publisher 5 est — — yes

HTML

sufficient

Kapadokya Galvaniz.

score 100 · 5,993 chars context · 1,967 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

insufficient

UNKNOWN

score 0 · 1,563 chars context · 537 in-tokens

Schema did not answer from 1563 chars of context — additional or different layer context needed.

AIPM

sufficient

The publisher is Kapadokya Galvaniz.

score 88 · 7,894 chars context · 2,406 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Q8 · Metadata · orientation

What is the content intent?

Gold: commercial

Gold source: contentIntent · html_independent

Sufficient: HTML, AIPM · Needs more / other context: Schema

Estimated retrieval cost · lowest cost AIPM · 3 input tokens

Layer Planner fields Input tokens Rounds Confidence OK?
HTML html.prose_signals 100 est — — yes
Schema No typed Schema field for this question. — est — — no
AIPM contentIntent 3 est — — yes

HTML

sufficient

The content intent is to promote Kapadokya Galvaniz's galvanizing services, particularly for cast parts, highlighting their quality, durability, and applications in various industries.

score 88 · 5,993 chars context · 1,966 in-tokens

HTML answered using 5993 chars of layer context (minimum fed this run).

Schema

insufficient

UNKNOWN.

score 0 · 1,563 chars context · 536 in-tokens

Schema did not answer from 1563 chars of context — additional or different layer context needed.

AIPM

sufficient

The content intent is commercial.

score 88 · 7,894 chars context · 2,405 in-tokens

AIPM answered using 7894 chars of layer context (minimum fed this run).

Methodology

  • AI Context Benchmark: Planner → Slice → LLM with measured API input tokens.
  • Reports describe sufficient layers, slice size, cost, and evidence — they do not declare a winning format.
  • Layers under test today: HTML, Schema.org, AIPM (extensible to RSS, Markdown, PDF, …).
  • Engine aipm_benchmark_v2 · Wed, Jul 29, 2026 9:06 AM · aipm_benchmark_score_v10

Share

Per-question context chain across layers.

Post on X LinkedIn WhatsApp
About the lab →