Intelligence

AI Platform

All systems operational

Intelligence

AI Platform Beta

Serve LLMs, embeddings, and vision models behind a unified API. One key, one bill, one network.

Tokens (30d)
4.2M
Inference calls
18.4K

+22% vs last 30 days

P95 latency
612ms

Within SLO (900 ms)

Spend (30d)
$84.20

42% of AI budget ($200)

Model catalog

Curated models available in EU Central · monthly flat-rate pricing per 1M tokens

All models
Most used

Llama 3.1 70B

Meta · instruction-tuned

General reasoning, long-context chat. Best for production workloads that need quality and control.

128k ctx ~640 ms
$0.72 / 1M in
New

Mistral Large 2

Mistral AI · EU-hosted

Fast, efficient European model. Excellent cost-to-quality for high-volume workloads.

32k ctx ~410 ms
$0.42 / 1M in

CLIP ViT-L/14

OpenAI · vision

Image understanding and zero-shot classification. Pairs well with text embeddings.

images ~120 ms
$0.18 / 1M img

BGE-M3 embeddings

BAAI · multilingual

Dense, sparse, and multi-vector retrieval. 100+ languages, ideal for RAG pipelines.

1024 dim ~28 ms
$0.06 / 1M tok

Whisper v3

OpenAI · speech-to-text

Robust transcription in 99 languages with automatic language detection.

audio ~340 ms
$0.30 / 1M audio-min
Beta

Llama 3.2 3B

Meta · edge-friendly

Small, fast, cheap. Ideal for classification, routing, and on-prem-style tasks.

128k ctx ~90 ms
$0.08 / 1M in

Playground

Test prompts against any deployed model before you wire it into your app.

Prompt 128 / 8192 tokens
Output completed in 612 ms
• Reuses physical connections, avoiding the per-request TLS + auth handshake cost. • Caps concurrent backends, protecting Postgres from connection storms under bursty traffic. • Reduces memory pressure on the database, since each backend has non-trivial RAM overhead.
0.7
512
0.95
612 ms · 412 tokens · $0.0003

Inference endpoint

POST https://ai.apisto.app/v1/chat/completions
Key sk-apisto-••••••••••••••••42f8
Auth
Bearer token
Rate limit
120 rpm · 40k tpm
Streaming
Supported

Usage by model

Last 30 days
Llama 3.1 · 70B 2.41M
Mistral Large 2 1.08M
BGE-M3 482k
CLIP ViT-L/14 142k
Whisper v3 84k

Recent requests

Logs
chat.completions · llama-3.1-70b checkout-api · 412 tok
642 ms 2 min ago
embeddings · bge-m3 web-frontend · 1.2k tok
48 ms 3 min ago
chat.completions · mistral-large-2 retried after 429 · analytics-worker
1.4 s 12 min ago
vision · clip-vit-l14 checkout-api · 1 image
118 ms 28 min ago
audio.transcriptions · whisper-v3 payload too large · 1.2 GB
— 1 hr ago

Quickstart

# Chat completion (streaming)
curl https://ai.apisto.app/v1/chat/completions \
  -H "Authorization: Bearer $APISTO_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-3.1-70b",
    "messages": [{"role": "user", "content": "Hello"}],
    "stream": true
  }'

Related

  • Prompt guardrails soon
  • Vector databases Open
  • Scoped AI keys Open