Ficelle
GitHub Install
No middleman · Open core · Free core forever

Free models break.
Your agents shouldn't.

When a free provider rate-limits or returns an error, Ficelle tries the next eligible free model in milliseconds. Paid fallback is blocked by design, so routed requests never incur paid-model charges.

Install in one command See it on GitHub
No account, no card No paid-model charges Your keys stay local
Routing example

Failover stays free, local, and controlled.

Core includes OpenRouter + Nous; Pro maintains the broader pool
OpenRouter Nous Mistral Groq Gemini NVIDIA Cerebras SiliconFlow Requesty OVH Cloud SambaNova Cloudflare Routeway
Explore Ficelle Pro
Why Ficelle

Free is easy. Free that works isn't.

When one model fails, Ficelle tries the next.

If a model rate-limits, returns errors, or fails a capability check, Ficelle excludes it and tries the next eligible option.

It stays free, guaranteed.

Ficelle only routes to models that are genuinely free and can run your tools. Nothing else qualifies.

It tests models so you don't have to.

Ficelle checks whether models handle tool calls, JSON, and vision, then continuously ranks working models by measured speed and reliability.

What to route

Not every agent task needs a frontier model.

In Ficelle's 76-task benchmark, free routing was quality-safe against gpt-5 for labelling, extraction, reasoning, and code. Route those workloads through Ficelle and keep a paid model for polished writing.

Labelling and triage

Tagging tickets, detecting intent, sorting inbound, picking a branch in an agent graph.

Route it free ficelle/auto-fast

Free tied the premium model on every task we tested.

Structured extraction

Pulling fields out of invoices, emails, scraped pages, logs. Anything that ends as JSON.

Route it free ficelle/auto-json

95% quality-safe in the benchmark.

Reasoning and code

Multi-step logic, small refactors, test scaffolding, working through a plan.

Route it free ficelle/auto-reasoning

The one we expected free to fail. It held up.

Verdicts come from a 76-task benchmark against gpt-5. See the numbers below →

$ http://127.0.0.1:8646/v1

Up and running in three steps.

  1. Point your tools at one endpoint

    Cursor, Codex, Continue, Open WebUI, and your own scripts can point at Ficelle's OpenAI-compatible endpoint. Hermes and OpenClaw use packaged connectors.

  2. Ask for the kind of model you need

    Call ficelle/auto-tools, auto-json, or auto-fast. You say what the job needs; Ficelle picks the real model that delivers it.

  3. Ficelle handles the failover

    It scores working free models, routes to the best available option, and tries another eligible model when one fails. You do not have to switch providers manually.

Drop-in

Use the OpenAI-compatible stack you already have.

Set one base URL and keep the client-side API key as a placeholder. Your real provider keys stay inside Ficelle, on your machine. OpenAI-compatible clients need no adapter; Hermes uses a packaged connector. 

from openai import OpenAI

client = OpenAI(
    base_url="http://127.0.0.1:8646/v1",
    api_key="ficelle-local",
)

client.chat.completions.create(
    model="ficelle/auto-tools",
    messages=[{"role": "user", "content": "..."}],
)
curl http://127.0.0.1:8646/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "ficelle/auto-json",
    "messages": [{"role": "user", "content": "..."}]
  }'
export OPENAI_BASE_URL="http://127.0.0.1:8646/v1"
export OPENAI_API_KEY="ficelle-local"

# Tools that read these OpenAI environment variables now
# route through Ficelle without code changes.
Benchmarked, not promised

We measured it: 76 real agent tasks, free vs gpt-5.

100%
classification
quality-safe vs gpt-5
95%
JSON extraction
quality-safe vs gpt-5
94%
reasoning & code
quality-safe vs gpt-5
~$61
estimated savings per 10k requests
at measured quality parity

The router is free. Pro does the upkeep for you.

The core router and its built-in profiles are free. Pro adds custom virtual models and maintains the broader provider pool alongside fusion, prompt compression, one-command updates and support.

Free
The routing core. Free forever.
0 €

Open core (BSL 1.1). No account, no card.

Install
Full routing + auto-fallback across free models
OpenRouter + Nous reference providers; bring your own keys
Local dashboard, built-in profiles & live benchmarks
Strict-zero: hardcoded & inspectable
Most popular
Pro
Ficelle maintains the provider integrations for you.
12 € 12 € 9.08 € /mo

Billed monthly. Cancel anytime. 109 € billed once a year.

Start 14-day trial
Full curated provider pool, fixes shipped when a provider breaks
Model fusion: a panel of models + a judge for answer synthesis
Native prompt compression to reduce tokens per call
Custom virtual models, exported to Hermes, OpenClaw & generic clients
One-command updates for the maintained provider pack
Priority support
Founder
Lifetime
Lifetime Pro. Limited to the first 200 licenses.
199 €

One payment. No renewals, ever.

Get the Founder license
Everything in Pro, forever
No renewals, ever
Founding-user status

14-day free trial, card required. Cancel anytime before it ends. The 199 € Founder license is capped to the first 200. Routing stays local; Pro periodically reconnects to refresh its entitlement.

The things people ask before installing.

Clear answers on cost, privacy, model quality, and what happens when a provider goes down.

Is it really free?

The open core is free under BSL 1.1. It includes strict-zero routing, OpenRouter and Nous reference providers, and the one-command installer. Pro adds the maintained provider pool, model fusion, prompt compression and one-command updates. You never pay for models through Ficelle, on any plan.

What happens when a provider rate-limits?

Ficelle marks the model as cooling down, tries the best working free alternative, and probes the limited one until it recovers. When another eligible model is available, your agent receives its response instead of the provider's 429.

Are free models good enough for real work?

For labelling, triage, structured extraction and most reasoning: yes. We benchmarked 76 real agent tasks against gpt-5 and free routing was quality-safe on classification (100%), JSON extraction (95%) and reasoning (94%). Polished long-form writing is the exception, where the paid model still wins. Route the first three through Ficelle and keep your budget for the writing.

Do my prompts or keys pass through your servers?

No. There is no middleman: the router runs on your machine, your API keys stay local, and prompts go straight to the provider you choose. Ficelle-operated servers receive only an optional update request and the Pro license check; set FICELLE_DISABLE_UPDATE_CHECK=1 to opt out of update checks. The license check carries license identity, never prompts or provider keys.

Can it accidentally spend my money?

No. Paid fallback is hardcoded off. It is not a setting you can mis-toggle, and Ficelle refuses to start if the config claims otherwise. Any model that reports billing gets quarantined automatically. The core code is public to inspect, so you can verify all of it.

Does the Pro trial need a card?

Yes. The 14-day trial requires a card. Stripe charges the selected Pro plan when the trial ends unless you cancel first. Annual costs 109 €/yr; Lifetime costs 199 € as a single payment with no renewals. Prices are in euros, and your bank converts the EUR amount at its usual card rate.

Which tools work with it?

Ficelle supports Cursor, Codex, Continue, Open WebUI, LangChain, CrewAI and plain curl through its OpenAI-compatible endpoint at http://127.0.0.1:8646/v1. Hermes and OpenClaw use packaged connectors. Claude Code cannot connect directly today, but its scripts and hooks can use Ficelle.

Can I create my own virtual models?

Ficelle Pro includes Profile Studio. You can create a stable ID such as ficelle/custom/coding, start from a built-in routing policy, tune its model pool and export it to Hermes, OpenClaw or a generic OpenAI-compatible client. If Pro expires, Ficelle preserves the configuration but stops listing, exporting and routing that custom ID; direct requests fail explicitly instead of silently falling back.

A note from the founder

I built Ficelle because my own agents kept breaking.

I run most of my own agent work on free models. Providers rate-limit, retire models, or quietly start reporting a price, and something breaks roughly every week. I got tired of finding out after the fact.

So I wrote a router that picks whatever works right now and reroutes the moment it stops. Then I made it structurally unable to spend my money, because the alternative to a broken agent should not be a surprise invoice.

Before charging for Pro, I ran 76 real tasks through free routing and gpt-5 to measure where free routing held up and where it did not. Those numbers are on this page, including the category where paid still wins.

Cyril Creator of Ficelle
Read the code

Keep your agents running.

One command. No account or card. Paid fallback is blocked by design.

$ curl -fsSL https://raw.githubusercontent.com/TheBlueHouse75/ficelle-open-core/v0.3.12/scripts/bootstrap-ficelle.py | python3 -

The command above installs the same version the footer advertises.