Call kimi-k2.8-preview through a single OpenAI-compatible endpoint — one key, no separate vendor account, no multi-provider wiring.
Moonshot AIChatReasoningCodingVisionPrices are per 1M tokens in USD · ¥6.6 = $1 · Billed in CNY
Kimi models target long-context reading and code generation, with dedicated coding variants tuned for repository-scale tasks.
At $0.98 per 1M input tokens it sits above the catalogue median of $0.61 — a capability-first choice. Best for coding agents and repository-scale edits. Every model here is reachable with the same API key, so trialling it against a stronger sibling costs one line of code.
from openai import OpenAI
client = OpenAI(
base_url="https://starseaapi.com/v1",
api_key="sk-your-key"
)
r = client.chat.completions.create(
model="kimi-k2.8-preview",
messages=[{"role": "user", "content": "Explain prefix caching in one paragraph"}]
)
print(r.choices[0].message.content)| Models | Input $/M | Output $/M | Intelligence index | Coding index |
|---|---|---|---|---|
| kimi-k2.6 | $0.98 | $4.09 | — | 61.8 |
| kimi-k2.7-code | $0.98 | $4.09 | — | 60.8 |
| kimi-k3 | $3.03 | $15.15 | 50.2 | 76.2 |