One API key for every
free
Unify 12 free-tier inference providers behind one gateway. OpenAI Chat Completions for Cursor and SDKs, Anthropic Messages for Claude Code — with effort-based routing, automatic failover, and usage analytics.
from openai import OpenAI
client = OpenAI(
base_url="https://api.apikeychain.dev/v1",
api_key="ak-•••••••••••••••",
)
resp = client.chat.completions.create(
model="keychain-high",
messages=[{"role": "user", "content": "Explain quantum tunneling."}],
)
print(resp.choices[0].message.content)What could you build?
Routing intelligence everywhere you call a model
Point any OpenAI-compatible client at your keychain URL. Pick an effort tier and let the router handle the rest.
Routing across the best free-tier inference networks
The platform
A control plane for free AI inference
Routing, failover, key management and observability — in one OpenAI-compatible gateway.
Effort-based routing
Pick fast, balanced, or best (`keychain-low` / `-medium` / `-high`). The router cascades through free models until one answers.
Claude Code ready
Native Anthropic Messages API with streaming, tools, and token counting — routed through your free-tier providers.
Automatic failover
A 429 or outage on one provider transparently rolls to the next.
Rate-limit cooldowns
Throttled providers are parked in a cooldown window until they recover.
Encrypted at rest
Every upstream provider key is sealed with authenticated encryption.
Unified keychain key
One revealable ak- key fronts everything. Bearer or x-api-key auth.
Bring your own models
Pin any model id into a tier, then reorder priority to taste.
Usage analytics
Per-model, per-provider request counts, token totals, and latency.
How it works
From twelve dashboards to one request
No SDK swaps, no per-provider glue. Point OpenAI or Anthropic clients at your keychain URL.
Connect your providers
Paste the free-tier keys you already have. They're encrypted the moment they arrive.
Pick an effort tier
Send `keychain-low`, `-medium`, or `-high` as the model — fast, balanced, or best free-tier cascades.
Ship — we handle the rest
Failover, cooldowns and retries happen server-side. You get a clean OpenAI response.
The catalog
Every free model, one key
All 69 free-tier models across 12 providers.
Google's multimodal flagship — fast 2.0 Flash, deep 2.5 Pro.
LPU inference — Llama 3.3 70B at hundreds of tokens/sec.
Wafer-scale speed for Llama- and Qwen-class open models.
Efficient European frontier models, open-weight roots.
Reasoning-first models that rival closed frontier labs.
A meta-gateway — dozens of free community models in one slot.
Open-source model hosting at production scale.
Enterprise-grade Command models, OpenAI-compatible.
NVIDIA-hosted open models via the NIM OpenAI-compatible API.
Enterprise-speed Llama inference on SambaNova Cloud.
Open models via the Hugging Face inference router.
Workers AI models on Cloudflare's global edge.
Routing tiers
Fast, balanced, or best — all free
Tiers trade speed for capability across free models, not price. Reorder, disable, or extend any cascade from your dashboard.
Smallest, fastest free models — autocomplete, classification, and high-volume calls.
- 1gemini-2.0-flash
- 2groq/llama-3.1-8b-instant
- 3cerebras/llama3.1-8b
- 4nim/meta/llama-3.1-8b-instruct
- 5sambanova/Meta-Llama-3.1-8B-Instruct
- 6openrouter/nvidia/llama-nemotron-nano-9b-v2:free
- 7openrouter/google/gemma-4-26b-a4b:free
- 8openrouter/openai/gpt-oss-20b:free
Everyday free models — solid quality and speed for chat and agents.
- 1gemini-2.0-flash
- 2groq/llama-3.3-70b-versatile
- 3mistral-small-latest
- 4nim/meta/llama-3.3-70b-instruct
- 5sambanova/Meta-Llama-3.3-70B-Instruct
- 6openrouter/google/gemma-4-31b:free
- 7openrouter/nvidia/nemotron-3-super:free
- 8openrouter/openai/gpt-oss-120b:free
Strongest free models — R1, Gemini 2.5 Pro, and Nemotron Ultra for hard problems.
- 1gemini-2.5-pro
- 2deepseek/deepseek-r1
- 3groq/llama-3.3-70b-versatile
- 4nim/nvidia/llama-3.1-nemotron-70b-instruct
- 5openrouter/nvidia/nemotron-3-ultra:free
- 6openrouter/poolside/laguna-m.1:free
- 7openrouter/tngtech/deepseek-r1t2-chimera:free
Quickstart
If you can call OpenAI or Claude, you're already done
Swap the base URL and key. Chat Completions, Responses API, and Anthropic Messages all route through the same cascade.
- Drop-in base URL & ak- key (Bearer or x-api-key)
- Server-side routing & failover
- Every call logged for analytics
# A different SDK + key per provider…
from google import genai
from groq import Groq
from mistralai import Mistral
g = genai.Client(api_key=GEMINI_KEY)
q = Groq(api_key=GROQ_KEY)
m = Mistral(api_key=MISTRAL_KEY)
# …and you hand-roll the failover.Interested in seeing how API Keychain works for your stack?
Spin up your unified key in under a minute and route across every free-tier model from a single endpoint.