Live market

kimi-k3

MoonshotAI: Kimi K3

Of list price: 10%
Output / 1M: 1.50 USDT
Sellers online: 5
Calls / min: 0

Calling this model

POST /chat/completions — the same request OpenAI's SDKs already send. Point the base URL at asale and change nothing else.

billed per token

INPUT

The conversation so far, oldest first.

Ceiling on the answer's length.

Penalise tokens by how often they already appeared.

Penalise tokens that appeared at all, pushing to new topics.

Stop generating when this text appears.

How long the model thinks before answering.

Send the answer as it is produced.

Output

Run it to see the answer here.

Or call it from your own code

curl https://gw.asale.ai/v1/chat/completions \
  -H "Authorization: Bearer $ASALE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3",
    "messages": [
      {
        "role": "user",
        "content": "Say hello in one short sentence."
      }
    ]
  }'

import os, requests

r = requests.post(
    "https://gw.asale.ai/v1/chat/completions",
    headers={"Authorization": f"Bearer {os.environ['ASALE_API_KEY']}"},
    json={
      "model": "kimi-k3",
      "messages": [
        {
          "role": "user",
          "content": "Say hello in one short sentence."
        }
      ]
    },
)
r.raise_for_status()
print(r.json())

const r = await fetch("https://gw.asale.ai/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.ASALE_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "kimi-k3",
    "messages": [
      {
        "role": "user",
        "content": "Say hello in one short sentence."
      }
    ]
  }),
});
if (!r.ok) throw new Error(await r.text());
console.log(await r.json());

Request fields

Field	Type	Accepts	Meaning
modelrequired	string	—	Which model serves the request.
messagesrequired	text	—	The conversation so far, oldest first.
max_tokens	int	— … 4096	Ceiling on the answer's length.
frequency_penalty	number	-2 … 2	Penalise tokens by how often they already appeared.
presence_penalty	number	-2 … 2	Penalise tokens that appeared at all, pushing to new topics.
stop	string	—	Stop generating when this text appears.
reasoning_effort	enum	none · minimal · low · medium · high · xhigh · max	How long the model thinks before answering.
stream	boolean	—	Send the answer as it is produced.

Response

{
  "id": "chatcmpl-…",
  "object": "chat.completion",
  "model": "kimi-k3",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "…"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 12,
    "total_tokens": 36
  }
}

Token type	List / 1M	Market / 1M	Of list price
Input	3.00	0.30	10%
Output	15.00	1.50	10%
Cache read	0.30	0.03	10%
Cache write	3.00	0.30	10%

Channel quality

What has been measured about the supply behind this price. Readings, not promises — and aggregate, because which seller failed is nobody's business but theirs and ours.

Online lanes: 5; 100 more asking above the market
Verified lanes: 3/3; Proved they serve this model
First byte (p50): 2.05s; p95 2.05s
Success rate: 96.0%; 124 settled turns, 7d
Sellable capacity: 280.0M; Tokens offered right now

Latency and verification are measured over the last day; the success rate comes from settled traffic. See the status page

Price trend

Currently selling at 10% of the vendor's list price. More sellers online push it down; more buyers push it back up to list.

Of list price: 10%
Sellers online: 5
Calls / min: 0

Market depth

Price / 1MSellers

15.00