Ref. 01 — Inference Routing

One API.
Every open model.

OpenAI-compatible inference routing for open models. Point any SDK at V01GA, pick a model, go — no per-model integrations, no infrastructure.

Get an API key →Read the docs

Uplink · v1Live

REQ/S1,204P50312msUPTIME99.97%

Quickstartcurl

curl https://sila-rhquln4is-silatech00s-projects.vercel.app/v1/chat/completions \
  -H "Authorization: Bearer v01ga_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "v01ga/qwen-3.8",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Works with the OpenAI SDK — set base_url to https://sila-rhquln4is-silatech00s-projects.vercel.app/v1

Available models

Qwen 3.8 27B

27B reasoning model. Free, EU-hosted (OVHcloud, France).

v01ga/qwen-3.8

Llama 3.3 70B

Meta's 70B instruction-tuned model. Free, EU-hosted.

v01ga/llama-3.3-70b

Mixtral 8x7B

Mistral's mixture-of-experts model. Free, EU-hosted.

v01ga/mixtral-8x7b

Mistral Nemo 12B

Mistral's 12B model with 128k context. Free, EU-hosted.

v01ga/mistral-nemo

Qwen 2.5 Coder 32B

32B code-specialized model. Free, EU-hosted.

v01ga/qwen-2.5-coder-32b

DeepSeek R1 Distill 32B

DeepSeek R1 distilled into 32B. Reasoning model. Free, EU-hosted.

v01ga/deepseek-r1-32b

Qwen 3 32B

Qwen's latest 32B model. Free, EU-hosted (Scaleway, France).

v01ga/qwen-3-32b

DeepSeek R1

Full DeepSeek R1 reasoning model. Free, EU-hosted.

v01ga/deepseek-r1

Gemma 3 27B

Google's 27B instruction-tuned model. Free, EU-hosted.

v01ga/gemma-3-27b

Ministral 3B

Mistral's lightweight 3B model. Free tier, EU-hosted.

v01ga/ministral-3b

Ministral 8B

Mistral's 8B model. Free tier, EU-hosted.

v01ga/ministral-8b

Mistral Small 3.1

Mistral's efficient small model. Free tier, EU-hosted.

v01ga/mistral-small

Codestral

Mistral's code-generation model. Free tier, EU-hosted.

v01ga/codestral

Qwen 3 30B A3B

Qwen's 30B active-parameter MoE model. Free, EU-hosted.

v01ga/qwen-3-30b-a3b

All free-tier during MVP. Rate-limited per key.

§ 01

OpenAI-compatible

Same request and response shape. Drop in any OpenAI SDK with no changes.

§ 02

Streaming built in

Server-Sent Events with optional usage metadata. Real-time token delivery.

§ 03

Free during MVP

Open models via Hugging Face. Fair-use rate limits. No credit card required.

One API.Every open model.

One API.
Every open model.