GetVibeAPIA high-performance model tuned for a broad range of tasks.
Prices in $, updated October 25, 2026.
Last 7 days: responses / client requests that reached a supplier. Retries of one request count once; input checks, disabled routes, and client cancellations do not lower model quality.
Availability
99.92%
client requests over 7 days
Average latency
4.42 s
to full response
TTFT
3.69 s
to first token
Speed
51 tok/s
generation speed, tok/s (median)
Kimi K2.8 from Moonshot AI (preview release): reasoning, tools and image input.
Comparison of public providers for this model: price, availability, TTFT, speed and cache. No traffic yet — available probes shown, otherwise «—».
kimi-k2.8:codebuddy-cn(default, suffix optional)Minimum charge per request — $0.0000143 (supplier rule): short requests are billed no less than this amount
| Provider | Input $/1M | Output $/1M | Context | Availability | TTFT | Speed | Cache hit | Caching |
|---|---|---|---|---|---|---|---|---|
Defaultdefault kimi-k2.8:codebuddy-cn(default, suffix optional)Minimum charge per request — $0.0000143 (supplier rule): short requests are billed no less than this amount | 0.038 | 0.16 | 500K / 32K | — | — | — | — | yes |
One line per price level: our input / output price per 1M tokens, the discount off the official list price, and how many connected merchant subscription accounts are offering there.
:codebuddyBased on 190 paid requests, last 24 hours
:codebuddy-cnBased on 338 paid requests, last 24 hours
POST https://getvibeapi.com/v1/chat/completionsPass your API key in the Bearer header. Keys start with sk-pv-.
Authorization: Bearer $GETVIBE_API_KEYResponses follow the Chat Completions contract: the text is in choices[0].message.content, token usage is in usage.
All examples use the public model address kimi-k2.8.
curl https://getvibeapi.com/v1/chat/completions \
-H "Authorization: Bearer $GETVIBE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "kimi-k2.8",
"messages": [
{
"role": "user",
"content": "Explain quantum computing in simple terms"
}
],
"stream": false
}'Fields are generated from the service's live OpenAPI schema.
modelmessagesstreammax_tokenstemperaturetop_preasoning_efforttoolstool_choiceresponse_format| Field | Type | Required | Description |
|---|---|---|---|
model | string | yes | Public model ID in vendor/model form (OpenRouter-style), e.g. anthropic/claude-opus-5. The bare claude-opus-5 spelling is accepted as an input alias and bills identically. model:alias pins a channel, e.g. openai/gpt-5.6-sol:yellow. GET /v1/models lists one ID per model by default; ?variants=1 returns the full list. |
messages | array | yes | No description |
stream | boolean | no | Stream the response as server-sent events (SSE)Default: false |
max_tokens | integer | no | Maximum number of tokens in the response |
temperature | number | no | Sampling randomness, 0–2Default: 1 |
top_p | number | no | No description |
reasoning_effort | string | no | Reasoning level for Chat Completions. The supported values depend on the model.none · minimal · low · medium · high · xhigh · max |
tools | array | no | List of tools (function calling) |
tool_choice | string | no | auto, none, required or a specific tool |
response_format | object | no | Force JSON output |
A high-performance model tuned for a broad range of tasks.
pip install openaifrom openai import OpenAI
client = OpenAI(
api_key="sk-pv-...",
base_url="https://getvibeapi.com/v1",
)
response = client.chat.completions.create(
model="kimi-k2.8",
messages=[{"role": "user", "content": "Explain quantum computing in simple terms"}],
)
print(response.choices[0].message.content)https://getvibeapi.com/v1kimi-k2.8Full documentation: getvibeapi.com/docs.
The live catalog contains 74 text models.
Kimi K2.8: 0.038 $ input and 0.16 $ output per 1M tokens.
The price of Kimi K2.8: $0.038 per 1M input · $0.16 per 1M output tokens. Pay only for the tokens you use — no subscriptions or flat fees.
Yes, the API is fully OpenAI-compatible. Just change base_url to https://getvibeapi.com/v1 and set your key — the rest of the code stays the same.