Statistics on AI × digital asset management.

Featured · Real research, cited sources

• Statistic · Vision API economics · New

  70×

  Cost spread between the cheapest and priciest frontier multimodal LLM, per 1 MP image

  Gemini 2.0 Flash $0.0002 to Claude Opus 4.7 (3 MP) $0.014. Computed from published per-token rates × documented image-token formulas. The "mini" models are not always the cheapest.

  AI Tagging·n=7 products

• Statistic · Third-party benchmark · New

  94%

  Top MMMU-Pro multimodal score, May 2026 (GPT-5.4 Pro)

  GPT-5.4 Pro 94% · Claude Mythos Preview 92.7% · Gemini 3.1 Pro 83.9%. 27 models scored on a benchmark designed to remove text shortcuts. Source: BenchLM.ai.

  AI Tagging·n=27 models

• Statistic · Benchmark methodology · New

  17–27pt drop

  Multimodal LLMs lose 17-27 points when text shortcuts are removed

  On MMMU-Pro's vision-only setting, Claude 3.5 Sonnet drops from 68.3% to 48.0%. Open-source models drop up to 42 points. Source: Yue et al., arXiv 2409.02813.

  AI Tagging·n=6 models cited

• Statistic · Market sizing · New

  $7.49B by 2032

  Image recognition API market: $3.12B (2025) to $7.49B (2032), 13.6% CAGR

  Three independent analysts (LP Information, Fortune Business Insights, Grand View Research) converge on a 13-15% CAGR. North America held 32.1% share in 2025.

  AI Tagging·3 sources

Field-tested · From our own work

• Statistic · Updated weekly

  1of 6

  DAM vendors shipping native MCP support

  As of May 2026, only one of six leading DAMs ships a native Model Context Protocol server. Field-tested.

  MCP & Integration·May 2026

• Statistic · Field-tested

  ~1 hrmedian

  Median DAM-to-LLM install time

  Across 4 architecture patterns. Range: under 2 minutes (MCP-native) to 6+ hours (webhook bridges).

  MCP & Integration·n=4 patterns

• Statistic · From the corpus

  5+tags / asset

  Average AI-generated tags per creative asset

  Across 1M+ creative assets. Includes objects, mood, brand elements, and clip-level video tags.

  AI Tagging·n=1M+ assets

• Statistic · Field-tested

  4patterns

  DAM-to-LLM architecture patterns we tested

  From MCP-native (fastest) to legacy webhook bridges (slowest). Each timed, screenshotted, scored.

  MCP & Integration·Multiple environments

• Statistic · Capability Index

  9of 10

  DAM vendors publishing public REST API docs

  Nine vendors publish REST docs reachable without a login. One gates documentation behind a trial signup.

  Vendors·n=10

• Statistic · Capability Index

  8of 10

  DAM vendors with documented webhook support

  Eight vendors document outgoing webhooks as a first-class integration surface. One offers partial; one has none.

  Vendors·n=10

• Statistic · Capability Index

  6of 10

  DAM vendors with documented AI features

  Six vendors publish technical AI documentation, not just marketing copy. Three publish partial docs; one publishes none.

  Vendors·n=10

• Statistic · Capability Index

  4of 10

  DAM vendors maintaining a public changelog

  Four maintain a versioned changelog. Three publish blog updates; three publish nothing observable.

  Vendors·n=10

• Statistic · Capability Index

  1of 10

  DAM vendors shipping a public GraphQL API

  Only one DAM vendor ships a public GraphQL endpoint. The others are REST-only.

  Vendors·n=10

• Statistic · AI Tagging Index

  3of 10

  Vision APIs with multimodal LLM reasoning

  Only three of ten leading image-tagging APIs can answer free-form natural-language questions about an image. The other seven return labels.

  AI Tagging·n=10

• Statistic · AI Tagging Index

  9of 10

  Vision APIs publishing per-unit pricing

  Nine of ten publish per-unit pricing publicly. One requires a sales call. LLM token pricing carries a 5-15× per-image cost premium over classical CV.

  AI Tagging·n=10

• Statistic · AI Tagging Index

  4of 10

  Vision APIs with a free tier (no credit card)

  Four of ten provide a free tier without a card. The three hyperscalers and all three frontier multimodal LLMs require payment.

  AI Tagging·n=10

• Statistic · AI Tagging Index

  5of 10

  Vision APIs with documented OCR / text extraction

  Five ship a documented OCR endpoint. The frontier LLMs do OCR via prompt, but it isn't a named capability.

  AI Tagging·n=10

• Statistic · AI Tagging Index

  6of 10

  Vision APIs with custom model training via API

  Six let operators train on their own labeled data via API. This is the column where classical CV still beats the frontier LLMs.

  AI Tagging·n=10

• Statistic · AI Tagging Index

  5of 10

  Vision APIs publishing a paid-tier SLA

  Five publish a paid-tier uptime SLA. Two of the three frontier multimodal LLMs publish only a status page, not a contract.

  AI Tagging·n=10

• Statistic · In preparation

  AI tagging accuracy (precision & recall)

  Precision/recall of LLM-generated creative tags across 10,000+ human-verified assets. From Report 05.

  AI Tagging·Q3 2026

• Statistic · In preparation

  Top performing creative tags by ROAS

  Which tags correlate with above-median ROAS across the corpus, cross-platform. From Report 02.

  Performance Creative·Q3 2026