getcompanion-ai/co-mono
1
0
Fork 0
mirror of https://github.com/getcompanion-ai/co-mono.git synced 2026-04-15 09:01:14 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Insert cache point on openrouter+anthropic completions (#584)

Browse source 
Co-authored-by: nathyong <nathyong@noreply.github.com>
This commit is contained in:
nathyong 2026-01-13 09:29:33 +11:00 • committed by GitHub
parent 8cb9a42067
commit 7b2c627079
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 35 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
packages/ai/CHANGELOG.md
View file

@ -5,6 +5,7 @@
### Added
- Added `serviceTier` option for OpenAI Responses requests ([#672](https://github.com/badlogic/pi-mono/pull/672) by [@markusylisiurunen](https://github.com/markusylisiurunen))
- **Anthropic caching on OpenRouter**: Interactions with Anthropic models via OpenRouter now set a 5-minute cache point using Anthropic-style `cache_control` breakpoints on the last assistant or user message. ([#584](https://github.com/badlogic/pi-mono/pull/584) by [@nathyong](https://github.com/nathyong))
## [0.44.0] - 2026-01-12

34
packages/ai/src/providers/openai-completions.ts
View file

@ -365,6 +365,7 @@ function createClient(model: Model<"openai-completions">, context: Context, apiK
function buildParams(model: Model<"openai-completions">, context: Context, options?: OpenAICompletionsOptions) {
const compat = getCompat(model);
const messages = convertMessages(model, context, compat);
maybeAddOpenRouterAnthropicCacheControl(model, messages);
const params: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming = {
model: model.id,
@ -410,6 +411,39 @@ function buildParams(model: Model<"openai-completions">, context: Context, optio
return params;
}
function maybeAddOpenRouterAnthropicCacheControl(
model: Model<"openai-completions">,
messages: ChatCompletionMessageParam[],
): void {
if (model.provider !== "openrouter" || !model.id.startsWith("anthropic/")) return;
// Anthropic-style caching requires cache_control on a text part. Add a breakpoint
// on the last user/assistant message (walking backwards until we find text content).
for (let i = messages.length - 1; i >= 0; i--) {
const msg = messages[i];
if (msg.role !== "user" && msg.role !== "assistant") continue;
const content = msg.content;
if (typeof content === "string") {
msg.content = [
Object.assign({ type: "text" as const, text: content }, { cache_control: { type: "ephemeral" } }),
];
return;
}
if (!Array.isArray(content)) continue;
// Find last text part and add cache_control
for (let j = content.length - 1; j >= 0; j--) {
const part = content[j];
if (part?.type === "text") {
Object.assign(part, { cache_control: { type: "ephemeral" } });
return;
}
}
}
}
function convertMessages(
model: Model<"openai-completions">,
context: Context,