DEVELOPER DOCS

KOPI TokenMax
One API, 50+ models

An OpenAI-compatible unified model gateway: one key, one endpoint, smart-routed across 50+ LLMs with automatic multi-upstream failover.

50+

Models connected

Failover chain

100%

OpenAI compatible

300 RPM

Default rate limit

Get started in 3 steps

Get an API key

Point your endpoint

Set base URL to https://kopiaiagent.com/v1 — any OpenAI-SDK-compatible client just works.

Pick a model & call

Recommended: kopi-o (daily) / kopi-flash (fastest) / kopi-o-pro (flagship). Switch per task anytime.

Models & pricing

Prices per million tokens (USD). “Metered” means real-time pricing — query /v1/pricing.

Model	Tier	Context	Input $/M	Output $/M	Notes
kopi-oRecommended	T2 Standard	262K	$0.30	$0.30	Daily driver · reasoning + multimodal
kopi-flash	T3 Fast	128K	$0.30	$0.30	Fastest · everyday chat
kopi-o-pro	T1 Flagship	1M	$0.30	$0.30	Flagship reasoning · strongest
kopi-o-flash	T3 Fast	262K	$0.80	$2.00	Fast & lightweight
kopi-qwen	T2 Standard	1M	$1.25	$5.00	Ultra-long context
kopi-kimi	T2 Standard	262K	$0.74	$2.50	Code specialist
kopi-gemini	T3 Fast	1M	$1.50	$5.00	Fast multimodal
kopi-gpt5	T1 Flagship	1M	$5.00	$15.00	Strongest closed-source reasoning
kopi-opus	T1 Flagship	1M	$7.00	$35.00	Deep analysis
kopi-ds-r1	T2 Standard	128K	Metered	Metered	Open reasoning · usage-based pricing

Code examples

Fully OpenAI Chat Completions compatible — just swap the base_url, no code changes.

curl · chat completions

curl https://bill.kopiagent.ai/v1/chat/completions \
  -H "Authorization: Bearer <your_api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kopi-o",
    "messages": [{"role": "user", "content": "Hello!"}],
    "max_tokens": 1024
  }'

python · openai sdk

from openai import OpenAI

client = OpenAI(
    api_key="<your_api_key>",
    base_url="https://bill.kopiagent.ai/v1",
)

response = client.chat.completions.create(
    model="kopi-o",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

curl · streaming (SSE)

curl https://bill.kopiagent.ai/v1/chat/completions \
  -H "Authorization: Bearer <your_api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kopi-flash",
    "messages": [{"role": "user", "content": "Hi"}],
    "stream": true
  }'

Smart routing & failover

TokenMax routes each request to the best-fit model; if an upstream times out, rate-limits or errors, it automatically falls through a 5-level chain — zero downtime, no client retries.

Smart routing — Picks the cheapest capable model per task, cutting LLM spend significantly.
5-level auto failover — Falls back on 4xx/5xx, timeouts, 429 rate limits and empty responses.
Quota & security — Keys stored as SHA256 hashes; per-client token quotas (402 over limit); 300 RPM default.

OpenAI compatibility matrix

Chat Completions✅ Fullmessages, stream, max_tokens, temperature, tools — all supported
Streaming✅ FullSSE format, ends with data: [DONE]
Vision✅ Supportedimage_url in content array
Function calling / Tools✅ Supportedtool_calls / function calling
Audio TTS✅ Available/v1/audio/speech (24kHz WAV), STT in progress
Image generation✅ Available/v1/images/generations
Embeddings⏳ Coming soonOn the roadmap

Get your API key

Start for free