DeepSeek, Kimi and GLM are live — one endpoint, one bill.›
Voyai

One endpoint.
Every open model.

Call DeepSeek, Kimi and GLM through a single OpenAI-compatible API. No contracts, no minimums — you pay for the tokens you actually use.

Start buildingView pricing

Free credit on sign-up. No card required.

Models
Get API key
from openai import OpenAI

client = OpenAI(
    base_url="https://api.voyai.net/v1",
    api_key="<API_KEY>",
)

response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V4-Flash",
    messages=[
        {"role": "user", "content": "Explain prompt caching in two sentences."},
    ],
)

print(response.choices[0].message.content)
Explain prompt caching in two sentences.
Sample response. Highlighted values are the only things you change.

Built for people shipping to production

Drop-in compatible

Fully OpenAI-compatible: chat completions, streaming, tool calling, JSON mode. Point your existing client at our base URL and keep the rest of your code.

Routed, not just proxied

Every model runs behind a pool of upstream accounts with health checks and automatic failover. When one provider degrades, your request is retried elsewhere before it ever fails.

Billing you can audit

Per-request token counts, cached-input discounts passed through, and a line item for every call. Set a spend limit per key so a runaway loop cannot drain your balance.

Models available today

Frontier open-weight models, priced per million tokens. No idle server costs.

Browse all models

DeepSeek V4 Flash

deepseek-ai/DeepSeek-V4-Flash

977K context
Tool callingJSON modeReasoningContext caching
Input
$0.5280
Output
$1.5840
Cached input
$0.0170

DeepSeek V4 Pro

deepseek-ai/DeepSeek-V4-Pro

977K context
Tool callingJSON modeReasoningContext caching
Input
$1.5840
Output
$4.7520
Cached input
$0.0530

GLM-5.3

zai-org/GLM-5.3

977K context
Tool callingJSON modeReasoningContext caching
Input
$1.3720
Output
$4.8000
Cached input
$0.3430

GLM-5.3 Flash

zai-org/GLM-5.3-Flash

977K context
Tool callingJSON modeReasoningVision
Input
$0.1370
Output
$0.4800
Cached input
$0.0400

Kimi K2.6

moonshotai/Kimi-K2.6

256K context
Tool callingJSON modeVisionContext caching
Input
$1.1150
Output
$4.6290
Cached input
$0.1880

DeepSeek V4 Flash Vision

deepseek-ai/DeepSeek-V4-Flash-Vision

977K context
Tool callingJSON modeReasoningVision
Input
$0.5280
Output
$1.5840
Cached input
$0.0170

You pay for tokens. Nothing else.

No seat fees, no monthly minimum, no charge for a model sitting idle. Cached input tokens are billed at a fraction of the normal rate and we pass that discount straight through.

See the full price list
{
  "usage": {
    "prompt_tokens": 1000,
    "prompt_cache_hit_tokens": 700,
    "prompt_cache_miss_tokens": 300,
    "completion_tokens": 50
  }
}
Cached input applies when a request reuses a prefix the upstream has already processed. It is billed at the cached rate automatically — nothing to enable.

Start with free credit

Create a key and make your first call in under a minute.

Create an account