# AKI Model Coverage Intelligence™ (MCI v1.0)

**Citation Object**: `aki:model-coverage:top100` | **Edition**: 2026.4.0 | **Updated**: 2026-10-11

> We measure AI model coverage. We verify the claims. We expose the gaps.

## Catalogue Leaderboards (Audited Directories & Providers)

| Rank | Platform | Category | Advertised | Measured | Variance | Identity Resolution | Trust Score | Status |
| :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :---
| #1 | [OpenRouter API Directory](https://openrouter.ai/models) | aggregator | 295 | **281** | -4.7% | 96% | **95.2** / 100 | Verified Fresh |
| #2 | [Artificial Analysis Directory](https://artificialanalysis.ai) | directory | 140 | **138** | -1.4% | 98% | **96.1** / 100 | Independently Audited |
| #3 | [Groq Cloud Models](https://console.groq.com/docs/models) | aggregator | 24 | **24** | 0.0% | 100% | **98.2** / 100 | Verified Fresh |
| #4 | [Together AI Library](https://together.ai/models) | aggregator | 185 | **180** | -2.7% | 94% | **92.0** / 100 | Verified Fresh |
| #5 | [Replicate Model Directory](https://replicate.com/explore) | aggregator | 420 | **395** | -5.9% | 92% | **91.5** / 100 | Documented |
| #6 | [Fireworks AI Platform](https://fireworks.ai/models) | aggregator | 110 | **106** | -3.6% | 95% | **93.2** / 100 | Verified Fresh |
| #7 | [DeepInfra Endpoints](https://deepinfra.com/models) | aggregator | 130 | **124** | -4.6% | 93% | **90.8** / 100 | Verified Fresh |
| #8 | [Ollama Model Library](https://ollama.com/library) | repo | 195 | **188** | -3.6% | 95% | **94.0** / 100 | Verified Fresh |
| #9 | [Hugging Face Open Hub](https://huggingface.co/models) | repo | 1,250,000 | **1,180,000** | -5.6% | 84% | **89.2** / 100 | Documented |
| #10 | [FreeLLM Index](https://freellm.sh) | directory | 900 | **742** | -17.5% | 78% | **81.8** / 100 | Discrepancy Flagged |
| #11 | [LLM Directory.org](https://llmdirectory.org) | directory | 650 | **520** | -20.0% | 76% | **80.0** / 100 | Discrepancy Flagged |
| #12 | [OpenAI Official Catalog](https://platform.openai.com/docs/models) | official | 42 | **42** | 0.0% | 100% | **99.2** / 100 | Verified Fresh |
| #13 | [Anthropic Official Catalog](https://docs.anthropic.com/en/docs/models-overview) | official | 18 | **18** | 0.0% | 100% | **98.6** / 100 | Verified Fresh |
| #14 | [Google Cloud Vertex Model Garden](https://cloud.google.com/vertex-ai/docs/model-garden) | official | 160 | **154** | -3.8% | 98% | **97.2** / 100 | Verified Fresh |
| #15 | [AWS Bedrock Foundation Catalog](https://aws.amazon.com/bedrock/models) | official | 95 | **92** | -3.2% | 97% | **96.5** / 100 | Verified Fresh |
| #16 | [Microsoft Azure AI Foundry](https://ai.azure.com/explore/models) | official | 1,800 | **1,720** | -4.4% | 96% | **95.5** / 100 | Verified Fresh |

## Canonical Model Identities

| Model | Developer | Family | Modalities | Release Date | Confidence | Access Status |
| :--- | :--- | :--- | :--- | :--- | :--- | :---
| [Claude 3.5 Sonnet](https://docs.anthropic.com/claude/sonnet) | Anthropic | Claude 3.5 | Text, Vision | 2024-06-20 | 100% | **Verified Access** |
| [GPT-4o (Omni)](https://platform.openai.com/docs/models/gpt-4o) | OpenAI | GPT-4 | Text, Vision, Audio | 2024-05-13 | 100% | **Verified Access** |
| [Llama 3.1 405B Instruct](https://ai.meta.com/llama) | Meta | Llama 3.1 | Text | 2024-07-23 | 98% | **Verified Access** |
| [Gemini 1.5 Pro](https://cloud.google.com/vertex-ai) | Google DeepMind | Gemini 1.5 | Text, Vision, Audio, Video | 2024-02-15 | 100% | **Verified Access** |
| [DeepSeek-V2.5](https://platform.deepseek.com) | DeepSeek | DeepSeek V2 | Text, Code | 2024-09-05 | 95% | **Verified Access** |
| [DeepSeek-V3](https://platform.deepseek.com) | DeepSeek | DeepSeek V3 | Text, Code, Reasoning | 2024-12-26 | 98% | **Verified Access** |
| [Grok 2 Beta](https://x.ai/api) | xAI | Grok 2 | Text, Vision | 2024-08-14 | 96% | **Verified Access** |
| [Mistral Large 2](https://docs.mistral.ai) | Mistral AI | Mistral Large | Text, Code | 2024-07-24 | 96% | **Verified Access** |
| [Qwen 2.5 72B Instruct](https://qwenlm.github.io) | Alibaba Cloud | Qwen 2.5 | Text, Code, Math | 2024-09-19 | 97% | **Verified Access** |
| [Command R+](https://docs.cohere.com) | Cohere | Command R | Text, Search, Tool | 2024-04-04 | 95% | **Verified Access** |
| [Claude 3.5 Haiku](https://docs.anthropic.com) | Anthropic | Claude 3.5 | Text, Vision | 2024-10-22 | 100% | **Verified Access** |
| [GPT-4o Mini](https://platform.openai.com) | OpenAI | GPT-4 | Text, Vision | 2024-07-18 | 100% | **Verified Access** |
| [Llama 3.2 11B Vision](https://ai.meta.com/llama) | Meta | Llama 3.2 | Text, Vision | 2024-09-25 | 98% | **Verified Access** |
| [Gemini 1.5 Flash](https://cloud.google.com/vertex-ai) | Google DeepMind | Gemini 1.5 | Text, Vision, Audio | 2024-05-14 | 100% | **Verified Access** |

## Claims Ledger (Advertised vs Measured)

| Audited Source | Claim Text | Claimed | Measured | Variance | Scope | Evidence |
| :--- | :--- | :--- | :--- | :--- | :--- | :---
| **FreeLLM Index** | Claims 900 active free AI models available across 58 providers | 900 | **742** | **-17.5%** | Free tier LLMs requiring no credit card | [Evidence](https://freellm.sh) |
| **OpenRouter API Directory** | Directory listing total available router models | 295 | **281** | **-4.7%** | Publicly routable endpoints with active health checks | [Evidence](https://openrouter.ai/models) |
| **Hugging Face Open Hub** | Total catalogued models including fine-tunes and checkpoint files | 1,250,000 | **1,180,000** | **-5.6%** | All repositories with valid model card metadata | [Evidence](https://huggingface.co/models) |
| **LLM Directory.org** | Claims complete index of 650 production LLM endpoints | 650 | **520** | **-20.0%** | Live production endpoints across global commercial APIs | [Evidence](https://llmdirectory.org) |
| **Replicate Model Directory** | Models discoverable via search catalog versus public API list | 420 | **395** | **-5.9%** | Public community and official model endpoints | [Evidence](https://replicate.com/explore) |
| **Together AI Library** | Available serverless and dedicated fine-tune model catalog | 185 | **180** | **-2.7%** | Active inference endpoints with public pricing | [Evidence](https://together.ai/models) |
| **Artificial Analysis Directory** | Comprehensive benchmarked foundation model coverage | 140 | **138** | **-1.4%** | Benchmarked models with verified quality scorecards | [Evidence](https://artificialanalysis.ai) |
| **DeepInfra Endpoints** | Active pay-as-you-go serverless models | 130 | **124** | **-4.6%** | Serverless endpoints with warm cluster availability | [Evidence](https://deepinfra.com/models) |
| **Ollama Model Library** | Official curated local model library tags | 195 | **188** | **-3.6%** | Curated model manifests supported by official CLI | [Evidence](https://ollama.com/library) |
| **Fireworks AI Platform** | Fast inference model catalog | 110 | **106** | **-3.6%** | Production ready serverless inference models | [Evidence](https://fireworks.ai/models) |
| **Google Cloud Vertex Model Garden** | Model Garden foundation model endpoints in Vertex AI | 160 | **154** | **-3.8%** | Managed foundation models available across all GCP regions | [Evidence](https://cloud.google.com/vertex-ai/docs/model-garden) |
| **Microsoft Azure AI Foundry** | Curated foundation model catalog in Azure AI Foundry | 1,800 | **1,720** | **-4.4%** | Model catalog items deployable via pay-as-you-go APIs | [Evidence](https://ai.azure.com/explore/models) |

## Coverage Gap Radar

| Missing Model | Omitted From | Category | Impact | Audit Note | Detected |
| :--- | :--- | :--- | :--- | :--- | :---
| **Claude 3.5 Sonnet (Direct Free Tier)** | FreeLLM Index | reasoning-frontier | **CRITICAL** | Platform markets frontier reasoning access but Sonnet tier requires external paid routing keys. | 2026-10-11 |
| **Gemini 1.5 Flash (Direct API)** | Together AI Library | multimodal-speed | **MODERATE** | Together AI catalog focuses on open-weights; closed-frontier Google endpoints omitted by design. | 2026-10-10 |
| **DeepSeek-Coder-V2-Lite-Base** | OpenRouter API Directory | specialist-code | **NICHE** | Base pre-trained checkpoint excluded in favor of Instruct-tuned endpoint. | 2026-10-11 |
| **Llama 3.1 405B FP8 (Full Precision)** | LLM Directory.org | open-weights-large | **CRITICAL** | Directory lists quantised 4-bit endpoints under unquantised full-weight names. | 2026-10-11 |
| **Grok 2 Vision Beta** | Replicate Model Directory | multimodal-frontier | **MODERATE** | xAI proprietary multimodal model unavailable in community container catalogue. | 2026-10-10 |
| **Command R+ (Tool Use Native)** | Fireworks AI Platform | enterprise-rag | **MODERATE** | Cohere flagship RAG model not available in serverless multi-tenant pool. | 2026-10-10 |
| **Mistral Large 2 (123B)** | DeepInfra Endpoints | reasoning-multilingual | **MODERATE** | Omitted from serverless pool due to 123B memory allocation limits. | 2026-10-11 |
| **Qwen 2.5 Coder 32B (Q8_0 Quant)** | Ollama Model Library | specialist-code | **NICHE** | Only Q4_K_M quant is default in library; Q8 requires custom Modelfile creation. | 2026-10-10 |
| **Llama 3.1 405B Full Weights** | Groq Cloud Models | open-weights-large | **CRITICAL** | SRAM cluster capacity constraints limit Groq deployment to 70B and 8B variants. | 2026-10-11 |
| **Claude 3.5 Haiku** | Artificial Analysis Directory | speed-frontier | **NICHE** | Latency benchmarks pending final laboratory token throughput stabilization. | 2026-10-11 |
| **DeepSeek-V3 Direct Endpoint** | AWS Bedrock Foundation Catalog | reasoning-open | **MODERATE** | AWS Bedrock model selection limited to US-approved commercial foundational labs. | 2026-10-11 |
| **Claude 3.5 Sonnet (Direct Azure)** | Microsoft Azure AI Foundry | frontier-code | **CRITICAL** | Available exclusively via Anthropic direct API or AWS Bedrock; not natively hosted in Azure. | 2026-10-10 |