Free LLM Hub — Analytical Report Auto-generated from data/hub.yaml and data/discovered_models.json Date: 2026-05-20
Executive Summary The free, cheap, and BYOK LLM API ecosystem is fragmenting fast. This report catalogs every option we track and surfaces what's actually accessible today.
Key numbers: - 188 entries curated in the hub - 14 distinct categories - 0 providers responding publicly to /v1/models - 0 models auto-discovered live
Top categories: - subscription-plan: 47 - inference-provider: 24 - specialty-api: 22 - provider-api: 18 - open-weights-family: 14 - inference-engine: 13 - gateway: 12 - vector-db: 8
Provider Breakdown by Category agent-framework (5) Name Pricing URL AutoGen {'model': 'self-hosted'} — CrewAI {'model': 'self-hosted'} — LangGraph {'model': 'self-hosted'} — Mastra {'model': 'self-hosted'} — Pydantic AI {'model': 'self-hosted'} —
Name Pricing URL Aider {'model': 'self-hosted'} — Cline {'model': 'self-hosted'} — Codex CLI {'model': 'self-hosted'} — Continue.dev {'model': 'self-hosted'} — OpenHands {'model': 'self-hosted'} —
desktop-ui (6) Name Pricing URL GPT4All {'model': 'self-hosted'} — Jan {'model': 'self-hosted'} — KoboldCpp {'model': 'self-hosted'} — LM Studio {'model': 'free'} — Open WebUI {'model': 'self-hosted'} — Text Generation WebUI {'model': 'self-hosted'} —
eval-framework (4) Name Pricing URL DeepEval {'model': 'self-hosted'} — OpenAI Evals {'model': 'self-hosted'} — Promptfoo {'model': 'self-hosted'} — Ragas {'model': 'self-hosted'} —
gateway (12) Name Pricing URL Arize Phoenix {'model': 'self-hosted'} — GitHub Models {'model': 'freemium', 'free_tier': True} — Helicone {'model': 'freemium'} — Kilo Code Gateway {'model': 'freemium', 'free_tier': True, 'requires_credit_card': False, 'trial_credits_usd': 5} — Langfuse {'model': 'freemium'} — LiteLLM {'model': 'self-hosted'} — NewAPI {'model': 'self-hosted'} — OneAPI {'model': 'self-hosted'} — OpenRouter {'model': 'freemium', 'free_tier': True} — Portkey AI Gateway {'model': 'freemium'} — RouteLLM {'model': 'self-hosted'} — Vercel AI Gateway {'model': 'freemium', 'free_tier': True} —
inference-engine (13) Name Pricing URL Aphrodite Engine {'model': 'self-hosted'} — CTranslate2 {'model': 'self-hosted'} — ExLlamaV2 {'model': 'self-hosted'} — KTransformers {'model': 'self-hosted'} — llama.cpp {'model': 'self-hosted'} — LMDeploy {'model': 'self-hosted'} — LocalAI {'model': 'self-hosted'} — MLC-LLM {'model': 'self-hosted'} — Ollama {'model': 'self-hosted'} — SGLang {'model': 'self-hosted'} — TensorRT-LLM {'model': 'self-hosted'} — Text Generation Inference (TGI) {'model': 'self-hosted'} — vLLM {'model': 'self-hosted'} —
inference-provider (24) Name Pricing URL Avian.io {'model': 'freemium', 'free_tier': True} — Baseten {'model': 'trial-credits', 'trial_credits_usd': 30} — Cerebras {'model': 'freemium', 'free_tier': True, 'requires_credit_card': False} — Chutes {'model': 'trial-credits'} — Cloudflare Workers AI {'model': 'freemium', 'free_tier': True} — DeepInfra {'model': 'pay-per-token', 'input_per_mtok_usd': 0.14, 'output_per_mtok_usd': 0.28} — Featherless AI {'model': 'subscription', 'monthly_usd': 10} — Fireworks AI {'model': 'trial-credits', 'trial_credits_usd': 1} — Groq {'model': 'freemium', 'free_tier': True, 'requires_credit_card': False} — HuggingFace Inference Providers {'model': 'freemium', 'free_tier': True, 'trial_credits_usd': 0.1} — Hyperbolic {'model': 'trial-credits'} — Kluster AI {'model': 'freemium', 'free_tier': True} — Lepton AI {'model': 'trial-credits', 'trial_credits_usd': 10} — LLM7.io {'model': 'free'} — Nebius {'model': 'trial-credits'} — Novita AI {'model': 'trial-credits'} — NVIDIA NIM {'model': 'freemium', 'free_tier': True} — Ollama Cloud {'model': 'freemium', 'free_tier': True} — OpenCode Zen {'model': 'freemium', 'free_tier': True} — SambaNova Cloud {'model': 'freemium', 'free_tier': True} — Scaleway Generative APIs {'model': 'freemium', 'free_tier': True} — SiliconFlow (硅基流动) {'model': 'freemium', 'free_tier': True} — Targon (Bittensor) {'model': 'freemium', 'free_tier': True} — Together AI {'model': 'trial-credits', 'trial_credits_usd': 5} —
llm-framework (6) Name Pricing URL DSPy {'model': 'self-hosted'} — Haystack {'model': 'self-hosted'} — LangChain {'model': 'self-hosted'} — LlamaIndex {'model': 'self-hosted'} — Semantic Kernel {'model': 'self-hosted'} — Vercel AI SDK {'model': 'free'} —
model-catalog (4) Name Pricing URL Civitai {'model': 'freemium'} — HuggingFace Hub {'model': 'freemium', 'free_tier': True} — models.dev {'model': 'free'} — ModelScope (Alibaba) {'model': 'freemium', 'free_tier': True} —
open-weights-family (14) Name Pricing URL DeepSeek — — Gemma (Google) — — GLM (Zhipu) — — gpt-oss (OpenAI) — — Granite (IBM) — — Hermes (Nous Research) — — InternLM — — Llama (Meta) — — Mistral / Mixtral — — OLMo (AllenAI) — — Phi (Microsoft) — — Qwen (Alibaba) — — SmolLM (HuggingFace) — — Yi (01.AI) — —
provider-api (18) Name Pricing URL 01.AI (Yi / 零一万物) {'model': 'trial-credits'} — Anthropic API {'model': 'pay-per-token', 'input_per_mtok_usd': 3.0, 'output_per_mtok_usd': 15.0} — Baidu Qianfan (ERNIE) {'model': 'freemium', 'free_tier': True} — Cohere {'model': 'freemium', 'free_tier': True} — DashScope (Alibaba) {'model': 'freemium', 'free_tier': True} — DeepSeek Platform {'model': 'pay-per-token', 'input_per_mtok_usd': 0.14, 'output_per_mtok_usd': 0.28} — Google AI Studio (Gemini) {'model': 'freemium', 'free_tier': True, 'requires_credit_card': False} — InternLM (Shanghai AI Lab) {'model': 'freemium', 'free_tier': True} — MiniMax {'model': 'trial-credits'} — Mistral AI (La Plateforme) {'model': 'freemium', 'free_tier': True} — Moonshot AI (Kimi) {'model': 'freemium', 'trial_credits_usd': None} — OpenAI API {'model': 'pay-per-token', 'input_per_mtok_usd': 1.25, 'output_per_mtok_usd': 10.0} — Perplexity Sonar API {'model': 'pay-per-token', 'trial_credits_usd': 5} — Reka AI {'model': 'trial-credits'} — StepFun (阶跃星辰) {'model': 'trial-credits', 'input_per_mtok_usd': 0.1, 'output_per_mtok_usd': 0.3} — Tencent Hunyuan {'model': 'freemium', 'free_tier': True} — xAI Grok API {'model': 'pay-per-token', 'input_per_mtok_usd': 3.0, 'output_per_mtok_usd': 15.0} — Zhipu AI (Z.AI / GLM) {'model': 'freemium', 'free_tier': True} —
specialty-api (22) Name Pricing URL AssemblyAI {'model': 'trial-credits', 'trial_credits_usd': 50} — AUTOMATIC1111 WebUI {'model': 'self-hosted'} — Black Forest Labs (FLUX) {'model': 'trial-credits'} — Cartesia (Sonic) {'model': 'freemium', 'free_tier': True} — ComfyUI {'model': 'self-hosted'} — Coqui XTTS {'model': 'self-hosted'} — Deepgram {'model': 'trial-credits', 'trial_credits_usd': 200} — ElevenLabs {'model': 'freemium', 'free_tier': True, 'monthly_usd': None} — fal.ai {'model': 'trial-credits'} — faster-whisper {'model': 'self-hosted'} — Jina AI {'model': 'freemium', 'free_tier': True} — Kling {'model': 'freemium', 'free_tier': True} — Kokoro TTS {'model': 'self-hosted'} — Mixedbread {'model': 'freemium', 'free_tier': True} — Nomic Atlas {'model': 'freemium', 'free_tier': True} — PlayHT {'model': 'freemium', 'free_tier': True} — Pollinations {'model': 'free'} — Replicate {'model': 'trial-credits', 'trial_credits_usd': 0.5} — Resemble AI {'model': 'trial-credits'} — Runway {'model': 'subscription', 'monthly_usd': 15} — Voyage AI {'model': 'freemium', 'free_tier': True} — Whisper {'model': 'self-hosted'} —
subscription-plan (47) Name Pricing URL Amazon Q Developer Pro {'model': 'subscription', 'monthly_usd': 19} — Anthropic Enterprise {'model': 'subscription', 'monthly_usd': None} — Augment Code Indie {'model': 'subscription', 'monthly_usd': 15} — AWS Bedrock {'model': 'pay-per-token'} — Azure OpenAI {'model': 'pay-per-token'} — ChatGPT Plus {'model': 'subscription', 'monthly_usd': 20} — ChatGPT Pro (new) {'model': 'subscription', 'monthly_usd': 100} — ChatGPT Pro (original) {'model': 'subscription', 'monthly_usd': 200} — Claude Code Pro {'model': 'subscription', 'monthly_usd': 20} — Claude Max 20x {'model': 'subscription', 'monthly_usd': 200} — Claude Max 5x {'model': 'subscription', 'monthly_usd': 100} — Claude Pro {'model': 'subscription', 'monthly_usd': 20} — Copy.ai Pro {'model': 'subscription', 'monthly_usd': 49} — Cursor Pro {'model': 'subscription', 'monthly_usd': 20} — Cursor Pro+ {'model': 'subscription', 'monthly_usd': 60} — Cursor Teams {'model': 'subscription', 'monthly_usd': 40} — Cursor Ultra {'model': 'subscription', 'monthly_usd': 200} — Descript Hobbyist {'model': 'subscription', 'monthly_usd': 12} — Devin (Cognition AI) {'model': 'subscription', 'monthly_usd': 500} — ElevenLabs Starter {'model': 'subscription', 'monthly_usd': 5} — Gemini Advanced {'model': 'subscription', 'monthly_usd': 20} — Gemini Ultra {'model': 'subscription', 'monthly_usd': 250} — GitHub Copilot Business {'model': 'subscription', 'monthly_usd': 19} — GitHub Copilot Enterprise {'model': 'subscription', 'monthly_usd': 39} — GitHub Copilot Pro {'model': 'subscription', 'monthly_usd': 10} —
… and 22 more.
vector-db (8) Name Pricing URL Chroma {'model': 'self-hosted'} — LanceDB {'model': 'self-hosted'} — Milvus {'model': 'self-hosted'} — pgvector {'model': 'self-hosted'} — Pinecone {'model': 'freemium', 'free_tier': True} — Qdrant {'model': 'freemium', 'free_tier': True} — Vespa {'model': 'self-hosted'} — Weaviate {'model': 'freemium', 'free_tier': True} — ## Auto-Discovered Models
(no discovery data yet)
Methodology Curation: Manual entries in data/hub.yaml reviewed for accuracy.Discovery: Async GET requests to public /v1/models endpoints with 8-second timeout and concurrency limit of 10. No credentials sent.Refresh cadence: Automated via GitHub Actions on every workflow run.Categories: free-tier, byok-gateway, self-hosted, trial, open-models, tool. How to contribute See https://github.com/felipetruman/free-llm-hub/blob/main/CONTRIBUTING.md (coming soon — Phase C of roadmap).
License MIT — data and scripts. Individual provider terms apply.