Claude w/ bedrock custom inference profile - no caching

[stevoland]

Description

With custom inference profile model.api.id == arn:aws:bedrock:xxx:yyy:application-inference-profile/zzz so the special case for anthropic caching isn't entered:

opencode/packages/opencode/src/provider/transform.ts

Lines 247 to 254 in d6caaee

	if (
	model.providerID === "anthropic" ||
	model.api.id.includes("anthropic") ||
	model.api.id.includes("claude") ||
	model.api.npm === "@ai-sdk/anthropic"
	) {
	msgs = applyCaching(msgs, model.providerID)
	}

and cached tokens always 0

tokens: {
    input: 19373,
    output: 12,
    reasoning: 0,
    cache: {
      write: 0,
      read: 0,
    },
  },

Propose this change

index f6b7ec8cbc..a0209b8c31 100644
--- a/packages/opencode/src/provider/transform.ts
+++ b/packages/opencode/src/provider/transform.ts
@@ -251,6 +251,8 @@
       model.providerID === "anthropic" ||
       model.api.id.includes("anthropic") ||
       model.api.id.includes("claude") ||
+      model.id.includes("anthropic") ||
+      model.id.includes("claude") ||
       model.api.npm === "@ai-sdk/anthropic"
     ) {
       msgs = applyCaching(msgs, model.providerID)

Which fixes for this config

"amazon-bedrock": {
  "options": {
    "profile": "xxx",
    "region": "yyy",
    "setCacheKey": true
  },
  "models": {
    "claude-sonnet-4.5": {
      "id": "arn:aws:bedrock:yyy:zzz:application-inference-profile/nnn"
    }
  }
},

Unless you have an alternative approach?

Will raise PR in the meantime, thanks!

Plugins

n/a

OpenCode version

1.1.26

Steps to reproduce

No response

Screenshot and/or share link

No response

Operating System

macOS 26.2

Terminal

ghostty

[alemairebe]

see #2746 for a solution

[stevoland]

@alemairebe thanks but I think I don't think that addresses the issue here. Specifically, missing cachePoint on messages leading to always 0 cache reads.

I may be missing something?

This is my proposal #9838