17-Facet AI & Agentic Capability Catalog
Separates model providers, weights, repositories, CPU/cloud runtimes, time-series foundation models, 3D/multimodal models, agent SDKs vs orchestrators vs managed services, 7-stage RAG, DSPy optimizers, MCP + A2A protocols, guardrails, tracing vs evaluation, memory policies vs vector/SQL substrates, and durable execution.
All 17 Facets (187) agent_development_orchestration ai_security_and_guardrails automation_ipaas_dag_durable embeddings_and_rerankers evaluation_and_regression_testing llm_frontier_cloud llm_open_weights_cloud memory_behavior_policies multimodal_vision_audio_3d observability_tracing_monitoring programmatic_prompt_workflow_optimization protocols_mcp_and_a2a rag_context_pipeline slm_cpu_edge_quantized team_simulation_provenance timeseries_foundation_models vector_and_transactional_storage
| Component & Publisher | Primary Facet | Capability Tags | Blast & SPDX | Deployment Modes |
|---|---|---|---|---|
| Phi-4-mini-instruct (Q4_K_M) Microsoft · model_family_version |
slm_cpu_edge_quantized |
cpu_inferencefunction_callingcompact_reasoning | A1 · MIT |
cpu_edge, cloud_api |
| Qwen3-Coder-7B-Instruct (GGUF) Alibaba Qwen · model_family_version |
slm_cpu_edge_quantized |
cpu_inferencecode_repairast_patching | A1 · Apache-2.0 |
cpu_edge, self_hosted |
| Meta Llama 3.2 3B Instruct (Q4) Meta · model_family_version |
slm_cpu_edge_quantized |
cpu_inferencescent_scoringrouter_tier1 | A1 · Llama-3.2 |
cpu_edge, mobile |
| Google Gemma 3 4B IT (Q4) Google · model_family_version |
slm_cpu_edge_quantized |
cpu_inferencevision_edgemultilingual | A1 · Gemma |
cpu_edge, self_hosted |
| BitNet b1.58 2B4T Microsoft Research · model_family_version |
slm_cpu_edge_quantized |
ternary_1_58bitultra_low_energy_cpu | A1 · MIT |
cpu_edge |
| SmolLM2-1.7B-Instruct Hugging Face · model_family_version |
slm_cpu_edge_quantized |
cpu_inferenceon_device_classifier | A1 · Apache-2.0 |
cpu_edge, browser_wasm |
| llama.cpp (GGML / AVX-512 / NEON) ggml-org · inference_runtime |
slm_cpu_edge_quantized |
cpu_simd_runtimegguf_quantizationgrammar_json | A1 · MIT |
cpu_edge, self_hosted |
| Intel OpenVINO GenAI Intel · inference_runtime |
slm_cpu_edge_quantized |
int4_cpu_npuxeon_amx | A1 · Apache-2.0 |
cpu_edge, self_hosted |
| ONNX Runtime GenAI Microsoft · inference_runtime |
slm_cpu_edge_quantized |
cross_platform_cpudirectml_npu | A1 · MIT |
cpu_edge, mobile |
| Qwen3-Coder-3B-Instruct (Q4_K_M) Alibaba Qwen · model_family_version |
slm_cpu_edge_quantized |
cpu_inferenceinline_diff_patch | A1 · Apache-2.0 |
cpu_edge |
| Qwen3-8B (GGUF Q4_K_M) Alibaba Qwen · model_family_version |
slm_cpu_edge_quantized |
cpu_inferencehybrid_thinking_toggle | A1 · Apache-2.0 |
cpu_edge, self_hosted |
| Mistral Small 3.1 24B (Q4) Mistral AI · model_family_version |
slm_cpu_edge_quantized |
workstation_cpu_gpuvision_json | A2 · Apache-2.0 |
cpu_edge, self_hosted |
| PyTorch ExecuTorch Meta · inference_runtime |
slm_cpu_edge_quantized |
mobile_arm_xnnpackwearable_bci_edge | A1 · BSD-3-Clause |
cpu_edge, mobile |
| DeepSeek-R1-Distill-Qwen-7B (Q4) DeepSeek · model_family_version |
slm_cpu_edge_quantized |
cpu_reasoningmath_verifier_prep | A1 · MIT |
cpu_edge, self_hosted |
| IBM Granite 3.2 8B Instruct (Q4) IBM · model_family_version |
slm_cpu_edge_quantized |
enterprise_apache2confidence_calibration | A1 · Apache-2.0 |
cpu_edge, self_hosted |
| MLC WebLLM (WebGPU / Wasm) MLC-AI · inference_runtime |
slm_cpu_edge_quantized |
zero_server_browser_inferencewasm_sandbox | A1 · Apache-2.0 |
cpu_edge, browser_wasm |