Every frontier model.
OneAPI.
ZenLLM routes requests across Claude, GPT, DeepSeek and more, with automatic failover and prepaid per-token pricing.
curl https://api.zenllm.org/v1/chat/completions \
-H "Authorization: Bearer $ZENLLM_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-fable-5",
"messages": [{"role": "user", "content": "Hello"}],
"stream": true
}'Routing
Fails over before
your user notices.
- 1Every model maps to an ordered provider chain
- 2Rate limits and dead streams trigger an instant retry on the next provider
- 3Failing providers cool down and recover on their own
POST/v1/chat/completionsclaude-fable-59.2s
provider-1429, cooldown 60s
provider-2200, streaming
provider-3priority 3
0.0sresolve chain: 3 providers for claude-fable-5
0.4sprovider-1 connect (priority 1)
2.1sprovider-1 responded 429, retry-after 60
2.1sprovider-1 marked degraded, cooldown 60s
2.2sprovider-2 connect (priority 2)
4.8sprovider-2 responded 200, stream open
4.9spiping SSE to client, same request id
5.0sRerouted mid-request. The client just saw tokens.
app.py
resp = client.chat.completions.create(
model="",
)
200streamed in 2.4s · same key, same code
Swap models in one line
Every model in the catalog answers on the same endpoint. Change the name, keep everything else.
dashboard · api keys
production300 rpmactive
staging60 rpmactive
Keys you control
Create, freeze and delete keys per app. Each one carries its own models, rate limit and expiry.
dashboard · usage
settling live
gpt-5.61,733 tok$0.0047
deepseek-v4-pro901 tok$0.0018
claude-fable-53,412 tok$0.0124
Spend you can watch
Token counts and cost land in your dashboard as each request finishes. No surprise invoices.
Pricing
The rates are the pricing.
Prepaid credits, metered per token. No subscriptions, no seat minimums, and credits never expire. What this table says is what your balance is charged.
| Model | Context | Input / 1M | Output / 1M | Cached / 1M |
|---|---|---|---|---|
Claude Fable 5 claude-fable-5 | 1M | $2.00 | $10.00 | $0.50 |
Claude Sonnet 5 claude-sonnet-5 | 200K | $3.00 | $15.00 | $0.30 |
GPT-5.6 gpt-5.6 | 400K | $1.25 | $10.00 | $0.13 |
GPT-5.4 Mini gpt-5.4-mini | 400K | $0.25 | $2.00 | $0.03 |
DeepSeek V4 Pro deepseek-v4-pro | 128K | $0.55 | $2.19 | $0.06 |
Kimi K2.6 kimi-k2.6 | 256K | $0.60 | $2.50 | $0.06 |
Questions, answered
Point your SDK at ZenLLM tonight.
Then delete the retry code you wrote during the last outage.