sscoble/pi-mono
1
0
Fork 0
mirror of https://github.com/badlogic/pi-mono.git synced 2026-11-05 23:37:25 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

feat(ai): use Chat Completions for the kimi-coding provider

Browse source 
Switch kimi-coding to OpenAI Chat Completions at api.kimi.com/coding/v1,
the protocol the official Kimi Code client uses. K3, K3-256K, and Kimi For
Coding accept tools declared in later system messages, so tools loaded
mid-conversation (e.g. by tool_search) keep the prompt cache. Highspeed
rejects tool-bearing system messages and keeps added tools top-level.

Re-lands b06c1f2 (reverted in 6a1f799b6) with MINIMUM_PI_VERSION bumped
to 1.2.1. Released clients implement only Anthropic Messages for
kimi-coding and ignore a model's api, so they must keep receiving the
last catalog published under 0.80.7.

closes #10720
This commit is contained in:
David Brailovsky 2026-10-15 15:29:11 +02:00
parent dfb03c3283
commit a3f265b1c6
10 changed files with 117 additions and 100 deletions
Download patch file Download diff file Expand all files Collapse all files

5
scripts/publish-model-catalog.mjs
View file

@ -25,7 +25,10 @@ import {
// The storage layout, index format, and version ordering are defined in
// model-catalog-protocol.ts, which pi.dev shares to serve these artifacts.
// Bump this only when generated model metadata requires behavior unavailable in older pi clients.
const MINIMUM_PI_VERSION = "0.80.7";
// Clients older than this keep the last catalog published under the previous minimum.
// 1.2.1: kimi-coding models use openai-completions at /coding/v1; older clients only
// implement anthropic-messages for kimi-coding and ignore a model's api.
const MINIMUM_PI_VERSION = "1.2.1";
const JSON_CONTENT_TYPE = "application/json; charset=utf-8";
const IMMUTABLE_CACHE_CONTROL = "public, max-age=31536000, immutable";
const INDEX_CACHE_CONTROL = "no-store";