# AgentAPIList

> Independent, source-linked comparison of AI agent and decision APIs: pricing, limits, status, and safety notes. Facts are dated (last verified 2026-10-16) and link to official sources; unverified items are labeled.

## APIs
- [OpenRouter Armor](https://agentapilist.com/apis/openrouter-armor/): New (beta, launched 2026-10-15): an OpenRouter plugin that checks every tool call an agent’s model makes against your plain-English rules and blocks out-of-policy calls before they reach your app. OpenRouter says it uses TypeSafe’s Jev decision model. Free during the beta.
- [OpenAI Decisions API](https://agentapilist.com/apis/openai-decisions/): Dedicated POST /v1/decisions endpoint that returns typed answers (probability, choice, or score) about text and images instead of generated text. Model: gpt-6-luna (only model available). Since Oct 15, 2026 GPT-Live can call it natively during voice sessions.
- [Microsoft-Decision-1](https://agentapilist.com/apis/microsoft-decision-1/): Small decision model in Microsoft Foundry that returns typed numeric decisions (noul / choice / score) about text or JSON state. No free-form output or rationale. Also callable through OpenRouter and Vercel AI Gateway without an Azure deployment.
- [Vercel AI Gateway — Decisions endpoint](https://agentapilist.com/apis/vercel-ai-gateway-decisions/): OpenAI-compatible POST /decisions endpoint on Vercel AI Gateway that routes to any decision model on the gateway (e.g. openai/gpt-6-luna-decisions, typesafe-ai/jev, microsoft/microsoft-decision-1).
- [Jev (TypeSafe AI)](https://agentapilist.com/apis/jev-typesafe-ai/): Hosted "System One" decision model from TypeSafe AI, and the model OpenRouter says powers its new OpenRouter Armor tool-call guard (beta, 2026-10-15). Send a state plus typed questions (noul, choice, score) to POST /v1/systemone and get probabilities back. jev-latest points to the stable jev-1.13.0; jev-1.14.0 is the preview (jev-preview).
- [Cloudflare Clef (Clef, Clef-flash, Clef-omni)](https://agentapilist.com/apis/cloudflare-clef/): Cloudflare-hosted decision models on Workers AI that take a state and a schema of typed questions (noul, choice, score) and return a probability for every allowed option. Three variants: Clef (27B), Clef-flash (9B), Clef-omni (30B MoE, 3B active).
- [Perplexity Decisions API (pplx-decider)](https://agentapilist.com/apis/perplexity-decisions/): Perplexity API endpoint (POST https://api.perplexity.ai/v1/decisions) that answers yes/no, multiple-choice, and scored questions about text, JSON, and images with probabilities. $0.02 per 1M input tokens on pplx-decider-v1.2 ($0.01 on the 4B Flash model), output free.
- [Inception Mercury Decide](https://agentapilist.com/apis/inception-mercury-decide/): Decision model served at POST /v1/decisions (model mercury-decide) that evaluates named questions against a shared state and returns typed answers with probabilities.
- [Liquid AI d1](https://agentapilist.com/apis/liquid-ai-d1/): Hosted decision model on the Liquid API (System One shape) returning noul, choice, and score answers with zero generated tokens; supports images. Open-weight variants also documented.
- [Strands Decider 2B (open source)](https://agentapilist.com/apis/strands-decider-2b/): Small open-source decision model (Qwen3.5-2B-Base + LoRA adapter + option-scoring head) for local routing, tool selection, guardrails, and evals. Apache-2.0, weights on Hugging Face.
- [AWS Bedrock Guardrails](https://agentapilist.com/apis/aws-bedrock-guardrails/): Configurable guardrails: content filters, denied topics, PII handling, contextual grounding, and Automated Reasoning checks; returns block/allow plus assessments.
- [Content Safety in Foundry Control Plane (formerly Azure AI Content Safety)](https://agentapilist.com/apis/azure-ai-content-safety/): Microsoft harm classifiers with severity levels, Prompt Shields, and groundedness detection, now marketed as Content Safety in Foundry Control Plane.
- [Gemini structured output (Vertex AI / Agent Platform)](https://agentapilist.com/apis/google-vertex-structured-output/): Constrain Gemini output to a JSON schema or a single enum value (response_mime_type text/x.enum) with response_schema. Google Cloud now documents this under the Gemini Enterprise Agent Platform name. No probabilities are documented on the structured-output page.
- [Anthropic Claude (tool use / structured outputs)](https://agentapilist.com/apis/anthropic-claude-structured-output/): JSON outputs (output_config.format) and strict tool use constrain Claude to a JSON schema via constrained decoding. Returns labels, not probabilities. Available via Anthropic, Bedrock, Google Cloud, and Microsoft Foundry.
- [OpenAI Moderation (omni-moderation)](https://agentapilist.com/apis/openai-moderation/): Free harm classifier returning per-category scores for text and images. Only the continuously upgraded omni-moderation-latest alias is documented.
- [Cohere Classify](https://agentapilist.com/apis/cohere-classify/): Few-shot / fine-tuned classification at POST /v1/classify returning per-label confidences. v1 path only; there is no v2 classify.
- [Laya (open source)](https://agentapilist.com/apis/laya/): Open-weight 421M encoder decision model (ModernBERT-large), Apache-2.0. laya-serve exposes a System One-style /v1/systemone endpoint; also runs in llama.cpp.
- [llama.cpp decision endpoint (/v1/systemone)](https://agentapilist.com/apis/llama-cpp-decision-endpoint/): llama-server in llama.cpp now serves a local System One-style /v1/systemone endpoint for open decision models.
- [DeepSeek Search & Fetch API](https://agentapilist.com/apis/deepseek-search-fetch/): Two paid web-access endpoints for agents and RAG: Search returns ranked results from DeepSeek’s own search engine, and Fetch turns one URL into Markdown. $0.75 per 1,000 successful calls on each, billed per call rather than per token. Opened to all API users on Oct 17, 2026.
- [Liquid AI d1-ultra](https://agentapilist.com/apis/liquid-ai-d1-ultra/): Liquid AI’s larger multimodal decision model, released Oct 15, 2026. It answers yes/no (noul), choice and score questions about text and images with calibrated probabilities and generates no tokens. $0.042 per 1M input tokens, output free. Top decision model on the Artificial Analysis Decision Index (65.2, checked 2026-10-17).
- [H2O-Lightning-4B (open source)](https://agentapilist.com/apis/h2o-lightning-4b/): Open-weight 4B decision model from H2O.ai (built on Qwen3.5-4B), Apache-2.0. Runs on stock vLLM 0.30.0 behind a small shim that serves /v1/systemone and an OpenAI-compatible /v1/decisions route. Text and up to 4 images per request.
- [LLM Gateway — Decisions](https://agentapilist.com/apis/llm-gateway-decisions/): Gateway with a System One endpoint (/v1/systemone, launched Sep 20, 2026) and an OpenAI-compatible /v1/decisions endpoint (added Oct 12, 2026) serving Jev, GPT-6 Luna Decisions, and Microsoft-Decision-1.

## Explainers
- [Decision model vs LLM](https://agentapilist.com/compare/decision-model-vs-llm/): Decision models (OpenAI Decisions, Microsoft-Decision-1, Jev, Perplexity pplx-decider) return probabilities for options you define and bill input tokens only; LLMs with structured output return a label or JSON with no documented probability. Verified prices, a 1M-ticket cost example, and when to use each.

## Comparisons
- [Interactive API comparison tool](https://agentapilist.com/compare/): pick 2–4 APIs and compare pricing, limits, auth, status and sources side by side; shareable as /compare/?apis=slug-a,slug-b.
- [OpenAI Decisions vs Jev](https://agentapilist.com/compare/openai-decisions-vs-jev/): As of 2026-10-15, Jev (TypeSafe AI) costs $0.042 per 1M input tokens versus $0.10 for the OpenAI Decisions API, publishes its context limits (64k/request, 262k long-context), and offers pinnable version IDs (jev-1.13.0 stable, jev-1.14.0 preview).
- [Microsoft-Decision-1 vs Jev](https://agentapilist.com/compare/microsoft-decision-1-vs-jev/): On 2026-10-15 both list $0.042 per 1M input tokens with free output, and both use the System One question types (noul, choice, score).
- [OpenAI Decisions vs Microsoft-Decision-1](https://agentapilist.com/compare/openai-decisions-vs-microsoft-decision-1/): As of 2026-10-15, Microsoft-Decision-1 is cheaper ($0.042 vs $0.10 per 1M input tokens) and publishes a 32,768-token context on OpenRouter, while the OpenAI Decisions API publishes no context or rate limits.
- [Jev vs Cloudflare Clef](https://agentapilist.com/compare/jev-vs-cloudflare-clef/): As of 2026-10-16, Cloudflare Clef-flash is slightly cheaper than Jev ($0.038 per 1M input tokens) but has a 24,576-token context; Jev costs $0.042 with 64k (262k long-context).
- [OpenAI Decisions vs Cloudflare Clef](https://agentapilist.com/compare/openai-decisions-vs-cloudflare-clef/): As of 2026-10-15, Cloudflare Clef-flash ($0.038/1M) undercuts the OpenAI Decisions API ($0.10/1M), while full Clef ($0.240/1M) costs more.

## Cost calculator
- [AI API cost calculator](https://agentapilist.com/cost-calculator/): monthly cost per API from requests, input and output tokens, cheapest first, with formula, price date and source; shareable as /cost-calculator/?r=100000&in=1000&out=300&cat=decision.
- [OpenAI Decisions API cost calculator](https://agentapilist.com/cost-calculator/openai-decisions/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Microsoft-Decision-1 cost calculator](https://agentapilist.com/cost-calculator/microsoft-decision-1/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Jev cost calculator](https://agentapilist.com/cost-calculator/jev-typesafe-ai/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Cloudflare Clef cost calculator](https://agentapilist.com/cost-calculator/cloudflare-clef/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Perplexity Decisions cost calculator](https://agentapilist.com/cost-calculator/perplexity-decisions/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Mercury Decide cost calculator](https://agentapilist.com/cost-calculator/inception-mercury-decide/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Liquid d1 cost calculator](https://agentapilist.com/cost-calculator/liquid-ai-d1/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Liquid d1-ultra cost calculator](https://agentapilist.com/cost-calculator/liquid-ai-d1-ultra/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [LLM Gateway Decisions cost calculator](https://agentapilist.com/cost-calculator/llm-gateway-decisions/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Bedrock Guardrails cost calculator](https://agentapilist.com/cost-calculator/aws-bedrock-guardrails/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Azure Content Safety cost calculator](https://agentapilist.com/cost-calculator/azure-ai-content-safety/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Gemini 3.8 Flash cost calculator](https://agentapilist.com/cost-calculator/google-vertex-structured-output/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Claude API cost calculator](https://agentapilist.com/cost-calculator/anthropic-claude-structured-output/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [OpenAI Decisions API vs Jev cost](https://agentapilist.com/cost-calculator/openai-decisions-vs-jev/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Microsoft-Decision-1 vs Jev cost](https://agentapilist.com/cost-calculator/microsoft-decision-1-vs-jev/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [OpenAI Decisions API vs Microsoft-Decision-1 cost](https://agentapilist.com/cost-calculator/openai-decisions-vs-microsoft-decision-1/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Jev vs Cloudflare Clef cost](https://agentapilist.com/cost-calculator/jev-vs-cloudflare-clef/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [OpenAI Decisions API vs Cloudflare Clef cost](https://agentapilist.com/cost-calculator/openai-decisions-vs-cloudflare-clef/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Bedrock Guardrails vs Azure Content Safety cost](https://agentapilist.com/cost-calculator/bedrock-guardrails-vs-azure-content-safety/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Gemini 3.8 Flash vs Claude API cost](https://agentapilist.com/cost-calculator/gemini-vs-claude-structured-output/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Perplexity Decisions vs Liquid d1 cost](https://agentapilist.com/cost-calculator/perplexity-decisions-vs-liquid-d1/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Mercury Decide vs Perplexity Decisions cost](https://agentapilist.com/cost-calculator/mercury-decide-vs-perplexity-decisions/): monthly cost at 10k–10M requests (prices verified 2026-10-17).
- [Liquid d1-ultra vs Perplexity Decisions cost](https://agentapilist.com/cost-calculator/liquid-d1-ultra-vs-perplexity-decisions/): monthly cost at 10k–10M requests (prices verified 2026-10-17).

## For agents
No key, CORS open, JSON responses. Same pricing engine as the calculator; prices verified 2026-10-17. Docs: https://agentapilist.com/for-agents/ · OpenAPI: https://agentapilist.com/openapi.json

- GET https://agentapilist.com/api/cost?requests=&input_tokens=&output_tokens=&text_units=&images=&apis=a,b&category= : monthly cost per API, cheapest first, with formula, price date, source URL, and not-comparable APIs with reasons. category is one of all, decision, structured, guardrail, classifier, gateway, search.
- POST https://agentapilist.com/mcp : remote MCP server (Streamable HTTP, JSON-RPC). Tools: estimate_cost, compare_apis, get_api, list_apis, list_agent_tools, get_agent_tool, compare_agent_tools, search.
- GET https://agentapilist.com/api/tools?free=&key_required=&mcp=&commercial=&category= : agent tools with filters (booleans true/false; category one of search, docs, reference, media, weather, calendar, finance, api-data). ?slug=x for one tool, ?compare=a,b for a feature grid.
- GET https://agentapilist.com/api/apis?category=&priced=&q= : tracked APIs (?slug=x for one).
- GET https://agentapilist.com/api/search?q=&type=all|apis|tools&limit= : keyword search across APIs and agent tools.
- Static data: https://agentapilist.com/data/pricing.json · https://agentapilist.com/data/apis.json · https://agentapilist.com/data/agent-tools.json

Examples:
```
curl "https://agentapilist.com/api/cost?requests=100000&input_tokens=1000&output_tokens=300&category=decision"
curl "https://agentapilist.com/api/cost?requests=1000000&input_tokens=500&output_tokens=5&apis=openai-decisions,jev-typesafe-ai"
curl "https://agentapilist.com/api/tools?key_required=false&mcp=true"
curl "https://agentapilist.com/api/search?q=weather"
curl -X POST https://agentapilist.com/mcp -H 'content-type: application/json' -H 'accept: application/json, text/event-stream' -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"estimate_cost","arguments":{"requests":100000,"input_tokens":1000,"output_tokens":300}}}'
```

## Free APIs for AI agents
- [Openverse API](https://agentapilist.com/agent-tools/openverse-api/): Search 5.2M+ openly licensed audio tracks and millions of images, with license data on every result. No key required.
- [Open-Meteo API](https://agentapilist.com/agent-tools/open-meteo-api/): Forecasts and current conditions for any latitude/longitude. No key for non-commercial use.
- [Wikipedia REST API](https://agentapilist.com/agent-tools/wikipedia-rest-api/): Article summaries and content from Wikipedia as JSON. No key; set a User-Agent.
- [You.com MCP (free keyless profile)](https://agentapilist.com/agent-tools/you-com-mcp-free/): Hosted MCP server with a keyless free profile: live web search for agents, capped at 100 queries per day.
- [Morpho MCP & API](https://agentapilist.com/agent-tools/morpho-mcp/): Read Morpho lending vaults, markets and positions from an agent. Hosted MCP and GraphQL, no key or account needed.
- [AlAdhan Prayer Times API](https://agentapilist.com/agent-tools/aladhan-prayer-times-api/): Daily prayer times and Hijri calendar dates for any city or coordinates. Free REST API, no key.
- [DexPaprika API + MCP](https://agentapilist.com/agent-tools/dexpaprika-api/): Live DEX pools, token prices and OHLCV across 36 chains. REST API and hosted MCP server, keyless to start (10,000 credits per IP per rolling 30 days).
- [Poly Haven API](https://agentapilist.com/agent-tools/poly-haven-api/): Search and download about 2,400 CC0 HDRIs, PBR textures and 3D models. Free for any use, no key; send a User-Agent naming your app.
- [GitBook docs MCP (per-site)](https://agentapilist.com/agent-tools/gitbook-docs-mcp/): Add {docs-site-url}/~gitbook/mcp to Claude, Cursor or Codex and the agent can search and read that product’s docs. Keyless for public sites.
- [AgentAPIList MCP server](https://agentapilist.com/agent-tools/agentapilist-mcp/): Keyless remote MCP server at https://agentapilist.com/mcp with 8 tools for AI API costs, API facts and free agent tools. This is our own tool.
- [AgentAPIList cost API + JSON data](https://agentapilist.com/agent-tools/agentapilist-cost-api/): GET /api/cost returns monthly cost per AI API with formula, price date and source; /data/*.json files carry the full price table. This is our own tool.

## About
- [All APIs](https://agentapilist.com/apis/)
- [Methodology](https://agentapilist.com/methodology/)
- [Sponsor](https://agentapilist.com/sponsor/)