Skip to content

Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)

Python

Report Chat cache hits in Anthropic usage (#1592)

Ali Khokhar11499f9c7bd7
diff --git a/pyproject.toml b/pyproject.toml
index ea33ca34..5aa8407f 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "5.15.2"
+version = "5.15.3"
description = "Local proxy connecting coding agents to OpenAI-compatible AI providers"
readme = "README.md"
requires-python = ">=3.14.0"
diff --git a/src/free_claude_code/providers/openai_chat/provider.py b/src/free_claude_code/providers/openai_chat/provider.py
index 545ff8cb..38a5afd4 100644
--- a/src/free_claude_code/providers/openai_chat/provider.py
+++ b/src/free_claude_code/providers/openai_chat/provider.py
@@ -719,8 +719,26 @@ class OpenAIChatProvider(BaseProvider):
return {}
def _anthropic_usage_fields(self, usage_info: Any) -> dict[str, int]:
- """Return provider-specific Anthropic usage fields for final SSE usage."""
- return {}
+ """Split standard cached prompt tokens for final Anthropic usage."""
+ prompt_tokens = usage_int(usage_info, "prompt_tokens")
+ cached_tokens = nested_usage_int(
+ usage_info,
+ "prompt_tokens_details",
+ "cached_tokens",
+ )
+ if (
+ prompt_tokens is None
+ or prompt_tokens < 0
+ or cached_tokens is None
+ or cached_tokens < 0
+ or cached_tokens > prompt_tokens
+ ):
+ return {}
+
+ return {
+ "input_tokens": prompt_tokens - cached_tokens,
+ "cache_read_input_tokens": cached_tokens,
+ }
async def _create_stream(
self,
diff --git a/tests/providers/test_openai_chat_usage.py b/tests/providers/test_openai_chat_usage.py
index 772fa544..9ff5cf05 100644
--- a/tests/providers/test_openai_chat_usage.py
+++ b/tests/providers/test_openai_chat_usage.py
@@ -7,9 +7,13 @@ from unittest.mock import AsyncMock, patch
import openai
import pytest
from httpx2 import Request, Response
+from openai.types.completion_usage import CompletionUsage, PromptTokensDetails
from free_claude_code.core.anthropic import ReasoningReplayMode
from free_claude_code.core.anthropic.models import MessagesRequest
+from free_claude_code.core.anthropic.sse_aggregation import (
+ aggregate_anthropic_sse_to_message,
+)
from free_claude_code.core.anthropic.stream_contracts import parse_sse_text
from free_claude_code.core.reasoning import DEFAULT_REASONING_POLICY, ReasoningPolicy
from free_claude_code.providers.admission import ProviderOperationKind
@@ -148,6 +152,110 @@ def test_usage_int_reads_dict_object_and_model_extra():
assert usage_int({"prompt_tokens": True}, "prompt_tokens") is None
+ ("usage", "expected"),
+ [
+ (
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": 15},
+ },
+ {"input_tokens": 7, "cache_read_input_tokens": 15},
+ ),
+ (
+ CompletionUsage(
+ completion_tokens=4,
+ prompt_tokens=22,
+ total_tokens=26,
+ prompt_tokens_details=PromptTokensDetails(cached_tokens=15),
+ ),
+ {"input_tokens": 7, "cache_read_input_tokens": 15},
+ ),
+ (
+ SimpleNamespace(
+ prompt_tokens=22,
+ prompt_tokens_details=None,
+ model_extra={
+ "prompt_tokens_details": {"cached_tokens": 15},
+ },
+ ),
+ {"input_tokens": 7, "cache_read_input_tokens": 15},
+ ),
+ (
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {"input_tokens": 22, "cache_read_input_tokens": 0},
+ ),
+ (
+ {
+ "prompt_tokens": 0,
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {"input_tokens": 0, "cache_read_input_tokens": 0},
+ ),
+ ],
+)
+def test_maps_standard_chat_cache_usage_to_anthropic_fields(usage, expected):
+ provider = _UsageTestProvider()
+
+ assert provider._anthropic_usage_fields(usage) == expected
+
+
+ "usage",
+ [
+ None,
+ {},
+ {"prompt_tokens_details": {"cached_tokens": 15}},
+ {"prompt_tokens": 22},
+ {"prompt_tokens": 22, "prompt_tokens_details": None},
+ {"prompt_tokens": 22, "prompt_tokens_details": "invalid"},
+ {
+ "prompt_tokens": -1,
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": -1},
+ },
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": 23},
+ },
+ {
+ "prompt_tokens": True,
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {
+ "prompt_tokens": 22.0,
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {
+ "prompt_tokens": "22",
+ "prompt_tokens_details": {"cached_tokens": 0},
+ },
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": True},
+ },
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": 15.0},
+ },
+ {
+ "prompt_tokens": 22,
+ "prompt_tokens_details": {"cached_tokens": "15"},
+ },
+ ],
+)
+def test_ignores_incomplete_or_inconsistent_standard_cache_usage(usage):
+ provider = _UsageTestProvider()
+
+ assert provider._anthropic_usage_fields(usage) == {}
+
+
def test_stream_usage_rejection_matches_usage_option_400():
error = _bad_request(
"Unrecognized request argument supplied: stream_options",
@@ -170,7 +278,12 @@ def test_stream_usage_rejection_does_not_match_unrelated_400():
async def test_openai_chat_stream_requests_usage_and_uses_provider_prompt_tokens():
provider = _UsageTestProvider()
request = make_messages_request(model="m")
- usage = SimpleNamespace(prompt_tokens=22, completion_tokens=4)
+ usage = CompletionUsage(
+ completion_tokens=4,
+ prompt_tokens=22,
+ total_tokens=26,
+ prompt_tokens_details=PromptTokensDetails(cached_tokens=15),
+ )
create = AsyncMock(
return_value=_stream(
[
@@ -200,7 +313,46 @@ async def test_openai_chat_stream_requests_usage_and_uses_provider_prompt_tokens
event.data["usage"] for event in parsed if event.event == "message_delta"
)
assert start_usage["input_tokens"] == 7
- assert final_usage == {"input_tokens": 22, "output_tokens": 4}
+ assert final_usage == {
+ "input_tokens": 7,
+ "output_tokens": 4,
+ "cache_read_input_tokens": 15,
+ }
+ assert sum(event.event == "message_delta" for event in parsed) == 1
+ assert sum(event.event == "message_stop" for event in parsed) == 1
+
+
+async def test_openai_chat_nonstream_message_uses_final_cache_partition():
+ provider = _UsageTestProvider()
+ request = make_messages_request(model="m")
+ usage = CompletionUsage(
+ completion_tokens=4,
+ prompt_tokens=22,
+ total_tokens=26,
+ prompt_tokens_details=PromptTokensDetails(cached_tokens=15),
+ )
+ create = AsyncMock(
+ return_value=_stream(
+ [
+ _chunk(content="hello"),
+ _chunk(finish_reason="stop"),
+ _chunk(usage=usage),
+ ]
+ )
+ )
+
+ with patch.object(provider._client.chat.completions, "create", create):
+ message, error = await aggregate_anthropic_sse_to_message(
+ provider.stream_messages(request, input_tokens=7)
+ )
+
+ assert error is None
+ assert message["usage"] == {
+ "input_tokens": 7,
+ "output_tokens": 4,
+ "cache_read_input_tokens": 15,
+ }
@pytest.mark.asyncio
diff --git a/uv.lock b/uv.lock
index 95b5f434..d040c576 100644
--- a/uv.lock
+++ b/uv.lock
@@ -607,7 +607,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "5.15.2"
+version = "5.15.3"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },