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

Fix Cloudflare Workers AI transport (#971)

Ali Khokharbd5157543035
diff --git a/.env.example b/.env.example
index dc3edae6..51fb5254 100644
--- a/.env.example
+++ b/.env.example
@@ -38,7 +38,7 @@ ZAI_API_KEY=""
FIREWORKS_API_KEY=""
-# Cloudflare AI REST Config (Anthropic-compatible Messages at api.cloudflare.com/client/v4/accounts/<id>/ai/v1)
+# Cloudflare Workers AI Config (OpenAI-compatible Chat Completions at api.cloudflare.com/client/v4/accounts/<id>/ai/v1)
CLOUDFLARE_API_TOKEN=""
CLOUDFLARE_ACCOUNT_ID=""
diff --git a/ARCHITECTURE.md b/ARCHITECTURE.md
index 26dc0373..44fb4afb 100644
--- a/ARCHITECTURE.md
+++ b/ARCHITECTURE.md
@@ -355,7 +355,10 @@ Gemini thought signatures, NIM tool-schema aliases and retry downgrades, or
DeepSeek attachment/tool/thinking compatibility. DeepSeek intentionally uses its
OpenAI-compatible Chat Completions endpoint because that is the endpoint that
reports prompt-cache hit/miss counters; the provider maps those counters back
-into Anthropic usage fields for Claude-compatible clients.
+into Anthropic usage fields for Claude-compatible clients. Cloudflare uses its
+account-scoped Workers AI OpenAI-compatible Chat Completions endpoint for
+`@cf/...` model IDs, while account ID composition, model search, and
+Cloudflare-specific reasoning deltas stay in the Cloudflare provider client.
NIM reasoning budget control is also treated as a provider-owned best-effort
downgrade: if an upstream NIM deployment rejects explicit budget control, FCC
retries without the budget while preserving thinking enablement.
diff --git a/README.md b/README.md
index 45020f85..054ee112 100644
--- a/README.md
+++ b/README.md
@@ -306,13 +306,13 @@ Fireworks exposes an **Anthropic-compatible** Messages API at `https://api.firew
Browse models at [fireworks.ai/models](https://fireworks.ai/models).
-### 14. [Cloudflare](https://developers.cloudflare.com/ai-gateway/)
+### 14. [Cloudflare](https://developers.cloudflare.com/workers-ai/)
Create a Cloudflare API token and copy your account ID from the Cloudflare dashboard.
-In the Admin UI, set `CLOUDFLARE_API_TOKEN` and `CLOUDFLARE_ACCOUNT_ID`, then set `MODEL` to a Cloudflare model slug such as `cloudflare/anthropic/claude-sonnet-4-5`.
+In the Admin UI, set `CLOUDFLARE_API_TOKEN` and `CLOUDFLARE_ACCOUNT_ID`, then set `MODEL` to a Cloudflare Workers AI model slug such as `cloudflare/@cf/moonshotai/kimi-k2.6`.
-This provider calls Cloudflare's account-scoped **Anthropic-compatible** Messages API at `https://api.cloudflare.com/client/v4/accounts/<account_id>/ai/v1/messages`. The first Cloudflare integration targets third-party Anthropic-compatible models; Cloudflare AI Gateway BYOK/provider-gateway routing is not required.
+This provider calls Cloudflare's account-scoped **OpenAI-compatible** Chat Completions API at `https://api.cloudflare.com/client/v4/accounts/<account_id>/ai/v1/chat/completions`. Use literal Workers AI model IDs, including the `@cf/` prefix when the catalog model includes it.
### 15. [Z.ai](https://z.ai/)
@@ -571,7 +571,7 @@ Important pieces:
- `fcc-codex` registers a custom `fcc` provider that points Codex at the local proxy's `/v1/responses` endpoint.
- Model routing resolves Claude model names to `MODEL_OPUS`, `MODEL_SONNET`, `MODEL_HAIKU`, or `MODEL`.
- NIM, DeepSeek, OpenCode Zen, and OpenCode Go use OpenAI chat streaming translated into Anthropic SSE.
-- Wafer, OpenRouter, Kimi, Fireworks AI, Cloudflare, Z.ai, LM Studio, llama.cpp, and Ollama use Anthropic Messages style transports where applicable (with provider-specific quirks and model-list URLs).
+- Wafer, OpenRouter, Kimi, Fireworks AI, Z.ai, LM Studio, llama.cpp, and Ollama use Anthropic Messages style transports where applicable (with provider-specific quirks and model-list URLs).
- The proxy normalizes thinking blocks, tool calls, token usage metadata, and provider errors into the shape each client expects.
- Request optimizations answer trivial Claude Code probes locally to save latency and quota.
diff --git a/config/provider_catalog.py b/config/provider_catalog.py
index 8bb9c9a8..cdaac316 100644
--- a/config/provider_catalog.py
+++ b/config/provider_catalog.py
@@ -214,7 +214,7 @@ PROVIDER_CATALOG: dict[str, ProviderDescriptor] = {
"cloudflare": ProviderDescriptor(
provider_id="cloudflare",
display_name="Cloudflare",
- transport_type="anthropic_messages",
+ transport_type="openai_chat",
credential_env="CLOUDFLARE_API_TOKEN",
credential_url="https://dash.cloudflare.com/profile/api-tokens",
credential_attr="cloudflare_api_token",
@@ -225,7 +225,6 @@ PROVIDER_CATALOG: dict[str, ProviderDescriptor] = {
"streaming",
"tools",
"thinking",
- "native_anthropic",
"rate_limit",
),
),
diff --git a/config/settings.py b/config/settings.py
index a463e57e..340ba94b 100644
--- a/config/settings.py
+++ b/config/settings.py
@@ -47,7 +47,7 @@ class Settings(BaseSettings):
# ==================== Fireworks AI Config ====================
fireworks_api_key: str = Field(default="", validation_alias="FIREWORKS_API_KEY")
- # ==================== Cloudflare AI REST Config ====================
+ # ==================== Cloudflare Workers AI Config ====================
cloudflare_api_token: str = Field(
default="", validation_alias="CLOUDFLARE_API_TOKEN"
)
diff --git a/providers/cloudflare/client.py b/providers/cloudflare/client.py
index b3c8af66..723584c7 100644
--- a/providers/cloudflare/client.py
+++ b/providers/cloudflare/client.py
@@ -1,7 +1,8 @@
-"""Cloudflare AI REST provider using Anthropic-compatible Messages."""
+"""Cloudflare Workers AI provider using OpenAI-compatible chat completions."""
from __future__ import annotations
+from collections.abc import Iterator, Mapping
from typing import Any
from urllib.parse import quote
@@ -9,25 +10,28 @@ import httpx
from providers.base import ProviderConfig
from providers.defaults import CLOUDFLARE_AI_REST_ROOT
-from providers.exceptions import AuthenticationError
-from providers.transports.anthropic_messages import (
- AnthropicMessagesTransport,
- NativeMessagesRequestPolicy,
- build_native_messages_request_body,
+from providers.exceptions import AuthenticationError, ModelListResponseError
+from providers.model_listing import (
+ ProviderModelInfo,
+ extract_openai_model_ids,
+ model_infos_from_ids,
+)
+from providers.transports.http import maybe_await_aclose
+from providers.transports.openai_chat import (
+ OpenAIChatRequestPolicy,
+ OpenAIChatTransport,
+ build_openai_chat_request_body,
)
-_ANTHROPIC_VERSION = "2023-06-01"
-_REQUEST_POLICY = NativeMessagesRequestPolicy(
+_REQUEST_POLICY = OpenAIChatRequestPolicy(
provider_name="CLOUDFLARE",
- extra_body="reject",
- reject_extra_body_message=(
- "Cloudflare native Messages API does not support extra_body on requests."
- ),
+ include_extra_body=True,
+ max_tokens_field="max_completion_tokens",
)
def cloudflare_ai_base_url(api_root: str | None, account_id: str) -> str:
- """Return the account-scoped Cloudflare AI REST base URL."""
+ """Return the account-scoped Cloudflare Workers AI OpenAI-compatible base URL."""
return f"{_cloudflare_account_api_url(api_root, account_id)}/ai/v1"
@@ -51,43 +55,105 @@ def _cloudflare_account_api_url(api_root: str | None, account_id: str) -> str:
return f"{root}/accounts/{encoded_account}"
-class CloudflareProvider(AnthropicMessagesTransport):
- """Cloudflare account-scoped AI REST provider."""
+class CloudflareProvider(OpenAIChatTransport):
+ """Cloudflare Workers AI OpenAI-compatible chat provider."""
def __init__(self, config: ProviderConfig, *, account_id: str):
base_url = cloudflare_ai_base_url(config.base_url, account_id)
self._model_search_url = _cloudflare_model_search_url(
config.base_url, account_id
)
+ self._model_list_client = httpx.AsyncClient(
+ proxy=config.proxy or None,
+ timeout=httpx.Timeout(
+ config.http_read_timeout,
+ connect=config.http_connect_timeout,
+ read=config.http_read_timeout,
+ write=config.http_write_timeout,
+ ),
+ )
super().__init__(
config.model_copy(update={"base_url": base_url}),
provider_name="CLOUDFLARE",
- default_base_url=base_url,
+ base_url=base_url,
+ api_key=config.api_key,
+ )
+
+ async def cleanup(self) -> None:
+ """Release provider client resources."""
+ await super().cleanup()
+ await self._model_list_client.aclose()
+
+ async def list_model_ids(self) -> frozenset[str]:
+ """Return Cloudflare Workers AI model ids from account model search."""
+ return frozenset(info.model_id for info in await self.list_model_infos())
+
+ async def list_model_infos(self) -> frozenset[ProviderModelInfo]:
+ """Return Cloudflare Workers AI model ids from account model search."""
+ response = await self._model_list_client.get(
+ self._model_search_url,
+ params={"format": "openrouter"},
+ headers=self._model_list_headers(),
)
+ try:
+ response.raise_for_status()
+ try:
+ payload = response.json()
+ except ValueError as exc:
+ raise ModelListResponseError(
+ "CLOUDFLARE model-list response is malformed: invalid JSON"
+ ) from exc
+ return model_infos_from_ids(
+ extract_openai_model_ids(payload, provider_name="CLOUDFLARE")
+ )
+ finally:
+ await maybe_await_aclose(response)
def _build_request_body(
self, request: Any, thinking_enabled: bool | None = None
) -> dict:
- return build_native_messages_request_body(
+ return build_openai_chat_request_body(
request,
thinking_enabled=self._is_thinking_enabled(request, thinking_enabled),
policy=_REQUEST_POLICY,
+ postprocessors=(_apply_cloudflare_request_quirks,),
)
- def _request_headers(self) -> dict[str, str]:
- return {
- "Accept": "text/event-stream",
- "Authorization": f"Bearer {self._api_key}",
- "Content-Type": "application/json",
- "anthropic-version": _ANTHROPIC_VERSION,
- }
+ def _handle_extra_reasoning(
+ self, delta: Any, ledger: Any, *, thinking_enabled: bool
+ ) -> Iterator[str]:
+ """Map Cloudflare's ``reasoning`` delta field to Anthropic thinking."""
+ reasoning = _cloudflare_reasoning(delta)
+ if not thinking_enabled or not reasoning:
+ return
+ yield from ledger.ensure_thinking_block()
+ yield ledger.emit_thinking_delta(reasoning)
def _model_list_headers(self) -> dict[str, str]:
return {"Authorization": f"Bearer {self._api_key}"}
- async def _send_model_list_request(self) -> httpx.Response:
- return await self._client.get(
- self._model_search_url,
- params={"format": "openrouter"},
- headers=self._model_list_headers(),
- )
+
+def _apply_cloudflare_request_quirks(
+ body: dict[str, Any], _request: Any, thinking_enabled: bool
+) -> None:
+ """Attach Cloudflare Workers AI chat-template thinking control."""
+ extra_body = body.setdefault("extra_body", {})
+ if not isinstance(extra_body, dict):
+ return
+ chat_template_kwargs = extra_body.setdefault("chat_template_kwargs", {})
+ if isinstance(chat_template_kwargs, dict):
+ chat_template_kwargs.setdefault("thinking", thinking_enabled)
+
+
+def _cloudflare_reasoning(delta: Any) -> str | None:
+ reasoning = getattr(delta, "reasoning", None)
+ if isinstance(reasoning, str) and reasoning:
+ return reasoning
+
+ model_extra = getattr(delta, "model_extra", None)
+ if isinstance(model_extra, Mapping):
+ reasoning = model_extra.get("reasoning")
+ if isinstance(reasoning, str) and reasoning:
+ return reasoning
+
+ return None
diff --git a/pyproject.toml b/pyproject.toml
index ff10a6d1..08413e6a 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "2.4.3"
+version = "2.4.4"
description = "Middleware between Claude Code CLI (Anthropic API) and NVIDIA NIM"
readme = "README.md"
requires-python = ">=3.14.0"
diff --git a/smoke/lib/config.py b/smoke/lib/config.py
index c428039a..d4b6772e 100644
--- a/smoke/lib/config.py
+++ b/smoke/lib/config.py
@@ -58,7 +58,7 @@ PROVIDER_SMOKE_DEFAULT_MODELS: dict[str, str] = {
"gemini": "gemini/models/gemini-3.1-flash-lite",
"groq": "groq/llama-3.3-70b-versatile",
"cerebras": "cerebras/llama3.1-8b",
- "cloudflare": "cloudflare/anthropic/claude-sonnet-4-5",
+ "cloudflare": "cloudflare/@cf/moonshotai/kimi-k2.6",
}
NVIDIA_NIM_CLI_DEFAULT_MODELS: tuple[str, ...] = (
diff --git a/tests/api/test_admin.py b/tests/api/test_admin.py
index b132cfab..bb98f585 100644
--- a/tests/api/test_admin.py
+++ b/tests/api/test_admin.py
@@ -298,7 +298,7 @@ def test_admin_apply_writes_cloudflare_fields_and_masks_preview(monkeypatch, tmp
"/admin/api/config/apply",
json={
"values": {
- "MODEL": "cloudflare/anthropic/claude-sonnet-4-5",
+ "MODEL": "cloudflare/@cf/moonshotai/kimi-k2.6",
"CLOUDFLARE_API_TOKEN": "cf-secret",
"CLOUDFLARE_ACCOUNT_ID": "cf-account",
}
@@ -312,7 +312,7 @@ def test_admin_apply_writes_cloudflare_fields_and_masks_preview(monkeypatch, tmp
assert "CLOUDFLARE_ACCOUNT_ID=cf-account" in body["env_preview"]
env_file = tmp_path / ".fcc" / ".env"
text = env_file.read_text(encoding="utf-8")
- assert "MODEL=cloudflare/anthropic/claude-sonnet-4-5" in text
+ assert "MODEL=cloudflare/@cf/moonshotai/kimi-k2.6" in text
assert "CLOUDFLARE_API_TOKEN=cf-secret" in text
assert "CLOUDFLARE_ACCOUNT_ID=cf-account" in text
diff --git a/tests/config/test_config.py b/tests/config/test_config.py
index 994ae84b..f0461352 100644
--- a/tests/config/test_config.py
+++ b/tests/config/test_config.py
@@ -684,8 +684,8 @@ class TestPerModelMapping:
({"MODEL": "deepseek/deepseek-chat"}, "deepseek/deepseek-chat", None),
({"MODEL": "wafer/DeepSeek-V4-Pro"}, "wafer/DeepSeek-V4-Pro", None),
(
- {"MODEL": "cloudflare/anthropic/claude-sonnet-4-5"},
- "cloudflare/anthropic/claude-sonnet-4-5",
+ {"MODEL": "cloudflare/@cf/moonshotai/kimi-k2.6"},
+ "cloudflare/@cf/moonshotai/kimi-k2.6",
None,
),
({"MODEL": "lmstudio/qwen2.5-7b"}, "lmstudio/qwen2.5-7b", None),
@@ -870,8 +870,7 @@ class TestPerModelMapping:
assert parse_provider_type("ollama/llama3.1") == "ollama"
assert parse_provider_type("wafer/DeepSeek-V4-Pro") == "wafer"
assert (
- parse_provider_type("cloudflare/anthropic/claude-sonnet-4-5")
- == "cloudflare"
+ parse_provider_type("cloudflare/@cf/moonshotai/kimi-k2.6") == "cloudflare"
)
assert parse_provider_type("gemini/models/gemini-3.1-flash-lite") == "gemini"
assert parse_provider_type("groq/llama-3.3-70b-versatile") == "groq"
@@ -893,8 +892,8 @@ class TestPerModelMapping:
assert parse_model_name("ollama/llama3.1") == "llama3.1"
assert parse_model_name("wafer/DeepSeek-V4-Pro") == "DeepSeek-V4-Pro"
assert (
- parse_model_name("cloudflare/anthropic/claude-sonnet-4-5")
- == "anthropic/claude-sonnet-4-5"
+ parse_model_name("cloudflare/@cf/moonshotai/kimi-k2.6")
+ == "@cf/moonshotai/kimi-k2.6"
)
assert (
parse_model_name("gemini/models/gemini-3.1-flash-lite")
diff --git a/tests/providers/test_cloudflare.py b/tests/providers/test_cloudflare.py
index 9a0897c1..a32ca216 100644
--- a/tests/providers/test_cloudflare.py
+++ b/tests/providers/test_cloudflare.py
@@ -1,7 +1,9 @@
-"""Tests for Cloudflare AI REST native Anthropic Messages provider."""
+"""Tests for Cloudflare Workers AI OpenAI-compatible chat provider."""
+from collections.abc import AsyncIterator
from contextlib import asynccontextmanager
-from unittest.mock import AsyncMock, MagicMock, patch
+from types import SimpleNamespace
+from unittest.mock import AsyncMock, patch
import httpx
import pytest
@@ -14,32 +16,7 @@ from providers.cloudflare import (
CloudflareProvider,
cloudflare_ai_base_url,
)
-from providers.exceptions import AuthenticationError, InvalidRequestError
-
-
-class FakeResponse:
- def __init__(self, *, status_code: int = 200, lines: list[str] | None = None):
- self.status_code = status_code
- self._lines = lines or []
- self.is_closed = False
- self.headers = httpx.Headers()
- self.request = httpx.Request("POST", f"{_BASE_URL}/messages")
-
- async def aiter_lines(self):
- for line in self._lines:
- yield line
-
- async def aiter_bytes(self, chunk_size: int = 65_536):
- if False:
- yield b""
-
- async def aclose(self):
- self.is_closed = True
-
- def raise_for_status(self):
- response = httpx.Response(self.status_code, request=self.request)
- response.raise_for_status()
-
+from providers.exceptions import AuthenticationError
_ACCOUNT_ID = "account-123"
_BASE_URL = f"{CLOUDFLARE_AI_REST_ROOT}/accounts/{_ACCOUNT_ID}/ai/v1"
@@ -63,9 +40,7 @@ def mock_rate_limiter():
async def _slot():
yield
- with patch(
- "providers.transports.anthropic_messages.transport.GlobalRateLimiter"
- ) as mock:
+ with patch("providers.transports.openai_chat.transport.GlobalRateLimiter") as mock:
instance = mock.get_scoped_instance.return_value
async def _passthrough(fn, *args, **kwargs):
@@ -81,7 +56,27 @@ def cloudflare_provider(cloudflare_config: ProviderConfig) -> CloudflareProvider
return CloudflareProvider(cloudflare_config, account_id=_ACCOUNT_ID)
-def test_cloudflare_ai_base_url_uses_account_scoped_messages_root() -> None:
+def _request(model: str = "@cf/moonshotai/kimi-k2.6") -> MessagesRequest:
+ return MessagesRequest(
+ model=model,
+ max_tokens=100,
+ messages=[Message(role="user", content="hi")],
+ )
+
+
+def _chunk(delta: SimpleNamespace, *, finish_reason: str = "stop") -> SimpleNamespace:
+ return SimpleNamespace(
+ choices=[SimpleNamespace(delta=delta, finish_reason=finish_reason)],
+ usage=SimpleNamespace(completion_tokens=5, prompt_tokens=8),
+ )
+
+
+async def _stream(*chunks: SimpleNamespace) -> AsyncIterator[SimpleNamespace]:
+ for chunk in chunks:
+ yield chunk
+
+
+def test_cloudflare_ai_base_url_uses_account_scoped_openai_chat_root() -> None:
assert cloudflare_ai_base_url(CLOUDFLARE_AI_REST_ROOT, "account/with slash") == (
f"{CLOUDFLARE_AI_REST_ROOT}/accounts/account%2Fwith%20slash/ai/v1"
)
@@ -94,80 +89,58 @@ def test_missing_account_id_raises_authentication_error(
CloudflareProvider(cloudflare_config, account_id=" ")
-def test_init_composes_account_scoped_base_url(
+def test_init_composes_account_scoped_openai_chat_base_url(
cloudflare_config: ProviderConfig,
) -> None:
- with patch("httpx.AsyncClient") as mock_client:
+ with (
+ patch("providers.transports.openai_chat.transport.AsyncOpenAI") as mock_openai,
+ patch("httpx.AsyncClient") as mock_httpx_client,
+ ):
provider = CloudflareProvider(cloudflare_config, account_id=_ACCOUNT_ID)
assert provider._api_key == "test-cloudflare-token"
assert provider._base_url == _BASE_URL
assert provider._model_search_url == _MODEL_SEARCH_URL
assert provider._provider_name == "CLOUDFLARE"
- assert mock_client.called
-
+ assert mock_openai.call_args.kwargs["base_url"] == _BASE_URL
+ assert mock_openai.call_args.kwargs["api_key"] == "test-cloudflare-token"
+ assert mock_httpx_client.called
-def test_request_headers_use_bearer_auth(cloudflare_provider: CloudflareProvider):
- headers = cloudflare_provider._request_headers()
- assert headers["Authorization"] == "Bearer test-cloudflare-token"
- assert headers["Accept"] == "text/event-stream"
- assert headers["Content-Type"] == "application/json"
- assert headers["anthropic-version"] == "2023-06-01"
- assert "x-api-key" not in headers
+def test_model_list_headers_use_bearer_auth(
+ cloudflare_provider: CloudflareProvider,
+) -> None:
assert cloudflare_provider._model_list_headers() == {
"Authorization": "Bearer test-cloudflare-token"
}
-def test_build_request_body_preserves_slash_model_id_and_forwards_thinking(
+def test_build_request_body_preserves_literal_cf_model_id_and_controls_thinking(
cloudflare_provider: CloudflareProvider,
) -> None:
request = MessagesRequest.model_validate(
{
- "model": "anthropic/claude-sonnet-4-5",
- "messages": [Message(role="user", content="Hello")],
+ "model": "@cf/moonshotai/kimi-k2.6",
+ "messages": [{"role": "user", "content": "Hello"}],
+ "max_tokens": 100,
"thinking": {"type": "enabled", "budget_tokens": 2048},
}
)
body = cloudflare_provider._build_request_body(request, thinking_enabled=True)
- assert body["model"] == "anthropic/claude-sonnet-4-5"
- assert body["stream"] is True
- assert body["thinking"] == {"type": "enabled", "budget_tokens": 2048}
-
-
-def test_build_request_body_strips_prior_thinking_blocks_when_disabled(
- cloudflare_provider: CloudflareProvider,
-) -> None:
- request = MessagesRequest.model_validate(
- {
- "model": "anthropic/claude-sonnet-4-5",
- "messages": [
- {
- "role": "assistant",
- "content": [
- {"type": "thinking", "thinking": "hidden"},
- {"type": "text", "text": "visible"},
- ],
- },
- {"role": "user", "content": "Continue"},
- ],
- }
- )
-
- body = cloudflare_provider._build_request_body(request, thinking_enabled=False)
+ assert body["model"] == "@cf/moonshotai/kimi-k2.6"
+ assert body["max_completion_tokens"] == 100
+ assert "max_tokens" not in body
+ assert body["extra_body"]["chat_template_kwargs"]["thinking"] is True
- assert body["messages"][0]["content"] == [{"type": "text", "text": "visible"}]
-
-def test_build_request_body_request_disabled_thinking_suppresses_thinking(
+def test_build_request_body_disabled_thinking_sets_cloudflare_template_flag(
cloudflare_provider: CloudflareProvider,
) -> None:
request = MessagesRequest.model_validate(
{
- "model": "anthropic/claude-sonnet-4-5",
+ "model": "@cf/moonshotai/kimi-k2.6",
"messages": [{"role": "user", "content": "Hello"}],
"thinking": {"type": "disabled"},
}
@@ -175,46 +148,51 @@ def test_build_request_body_request_disabled_thinking_suppresses_thinking(
body = cloudflare_provider._build_request_body(request, thinking_enabled=True)
- assert "thinking" not in body
+ assert body["extra_body"]["chat_template_kwargs"]["thinking"] is False
-def test_build_request_body_rejects_extra_body(
+def test_build_request_body_preserves_user_extra_body_without_overriding_thinking(
cloudflare_provider: CloudflareProvider,
) -> None:
request = MessagesRequest.model_validate(
{
- "model": "anthropic/claude-sonnet-4-5",
+ "model": "@cf/moonshotai/kimi-k2.6",
"messages": [{"role": "user", "content": "Hello"}],
- "extra_body": {"custom": True},
+ "extra_body": {"chat_template_kwargs": {"thinking": False}},
}
)
- with pytest.raises(InvalidRequestError, match="does not support extra_body"):
- cloudflare_provider._build_request_body(request)
+ body = cloudflare_provider._build_request_body(request, thinking_enabled=True)
+
+ assert body["extra_body"]["chat_template_kwargs"]["thinking"] is False
@pytest.mark.asyncio
async def test_lists_models_from_cloudflare_model_search_endpoint(
cloudflare_provider: CloudflareProvider,
) -> None:
+ response = httpx.Response(
+ 200,
+ json={
+ "object": "list",
+ "data": [
+ {"id": "@cf/moonshotai/kimi-k2.6", "object": "model"},
+ {"id": "@cf/meta/llama-4-scout-17b-16e-instruct", "object": "model"},
+ ],
+ },
+ request=httpx.Request("GET", _MODEL_SEARCH_URL),
+ )
with patch.object(
- cloudflare_provider._client,
+ cloudflare_provider._model_list_client,
"get",
new_callable=AsyncMock,
- return_value=httpx.Response(
- 200,
- json={
- "object": "list",
- "data": [
- {"id": "anthropic/claude-sonnet-4-5", "object": "model"},
- {"id": "anthropic/claude-opus-4-5", "object": "model"},
- ],
- },
- request=httpx.Request("GET", _MODEL_SEARCH_URL),
- ),
+ return_value=response,
) as mock_get:
assert await cloudflare_provider.list_model_ids() == frozenset(
- {"anthropic/claude-sonnet-4-5", "anthropic/claude-opus-4-5"}
+ {
+ "@cf/moonshotai/kimi-k2.6",
+ "@cf/meta/llama-4-scout-17b-16e-instruct",
+ }
)
mock_get.assert_awaited_once_with(
@@ -225,43 +203,115 @@ async def test_lists_models_from_cloudflare_model_search_endpoint(
@pytest.mark.asyncio
-async def test_stream_uses_post_messages_path(
+async def test_stream_uses_openai_chat_completions(
cloudflare_provider: CloudflareProvider,
) -> None:
- request = MessagesRequest(
- model="anthropic/claude-sonnet-4-5",
- messages=[Message(role="user", content="hi")],
+ delta = SimpleNamespace(
+ content="Hello from Cloudflare",
+ reasoning_content=None,
+ reasoning=None,
+ tool_calls=None,
)
- response = FakeResponse(
- lines=[
- "event: message_start",
- 'data: {"type":"message_start"}',
- "",
- "event: message_stop",
- 'data: {"type":"message_stop"}',
- "",
+
+ with patch.object(
+ cloudflare_provider._client.chat.completions,
+ "create",
+ new_callable=AsyncMock,
+ return_value=_stream(_chunk(delta)),
+ ) as mock_create:
+ events = [
+ event async for event in cloudflare_provider.stream_response(_request())
]
+
+ parsed = parse_sse_text("".join(events))
+ assert any(
+ event.event == "content_block_delta"
+ and event.data.get("delta", {}).get("text") == "Hello from Cloudflare"
+ for event in parsed
)
+ assert mock_create.call_args.kwargs["model"] == "@cf/moonshotai/kimi-k2.6"
+ assert mock_create.call_args.kwargs["stream"] is True
- with (
- patch.object(
- cloudflare_provider._client, "build_request", return_value=MagicMock()
- ) as mock_build,
- patch.object(
- cloudflare_provider._client,
- "send",
- new_callable=AsyncMock,
- return_value=response,
- ),
+
+async def test_stream_maps_cloudflare_reasoning_delta_to_thinking(
+ cloudflare_provider: CloudflareProvider,
+) -> None:
+ delta = SimpleNamespace(
+ content=None,
+ reasoning_content=None,
+ reasoning="Cloudflare reasoning",
+ tool_calls=None,
+ )
+
+ with patch.object(
+ cloudflare_provider._client.chat.completions,
+ "create",
+ new_callable=AsyncMock,
+ return_value=_stream(_chunk(delta)),
+ ):
+ events = [
+ event async for event in cloudflare_provider.stream_response(_request())
+ ]
+
+ parsed = parse_sse_text("".join(events))
+ assert any(
+ event.event == "content_block_delta"
+ and event.data.get("delta", {}).get("thinking") == "Cloudflare reasoning"
+ for event in parsed
+ )
+
+
+async def test_stream_maps_openai_tool_calls_to_tool_use(
+ cloudflare_provider: CloudflareProvider,
+) -> None:
+ tool_call = SimpleNamespace(
+ index=0,
+ id="call_1",
+ function=SimpleNamespace(name="echo", arguments='{"value":"x"}'),
+ )
+ delta = SimpleNamespace(
+ content=None,
+ reasoning_content=None,
+ reasoning=None,
+ tool_calls=[tool_call],
+ )
+ request = MessagesRequest.model_validate(
+ {
+ "model": "@cf/moonshotai/kimi-k2.6",
+ "messages": [{"role": "user", "content": "Use the tool"}],
+ "tools": [
+ {
+ "name": "echo",
+ "description": "Echo a value",
+ "input_schema": {
+ "type": "object",
+ "properties": {"value": {"type": "string"}},
+ "required": ["value"],
+ },
+ }
+ ],
+ }
+ )
+
+ with patch.object(
+ cloudflare_provider._client.chat.completions,
+ "create",
+ new_callable=AsyncMock,
+ return_value=_stream(_chunk(delta, finish_reason="tool_calls")),
):
events = [event async for event in cloudflare_provider.stream_response(request)]
- assert [event.event for event in parse_sse_text("".join(events))] == [
- "message_start",
- "message_stop",
- ]
- assert response.is_closed
- assert mock_build.call_args.args[:2] == ("POST", "/messages")
- assert mock_build.call_args.kwargs["headers"]["Authorization"] == (
- "Bearer test-cloudflare-token"
+ parsed = parse_sse_text("".join(events))
+ assert any(
+ event.event == "content_block_start"
+ and event.data.get("content_block", {}).get("type") == "tool_use"
+ and event.data.get("content_block", {}).get("name") == "echo"
+ for event in parsed
+ )
+ assert any(
+ event.event == "content_block_delta"
+ and event.data.get("delta", {}).get("partial_json") == '{"value":"x"}'
+ for event in parsed
)
diff --git a/tests/providers/test_provider_runtime.py b/tests/providers/test_provider_runtime.py
index 71e07e28..94f0f3a1 100644
--- a/tests/providers/test_provider_runtime.py
+++ b/tests/providers/test_provider_runtime.py
@@ -136,10 +136,10 @@ def test_zai_provider_config_ignores_stale_base_url_setting():
def test_cloudflare_descriptor_uses_api_root_not_account_url():
descriptor = PROVIDER_CATALOG["cloudflare"]
- assert descriptor.transport_type == "anthropic_messages"
+ assert descriptor.transport_type == "openai_chat"
assert descriptor.default_base_url == "https://api.cloudflare.com/client/v4"
assert descriptor.base_url_attr is None
- assert "native_anthropic" in descriptor.capabilities
+ assert "native_anthropic" not in descriptor.capabilities
assert "thinking" in descriptor.capabilities
@@ -149,7 +149,7 @@ def test_create_cloudflare_provider_uses_account_scoped_base_url():
cloudflare_account_id="test-account",
)
- with patch("httpx.AsyncClient"):
+ with patch("providers.transports.openai_chat.transport.AsyncOpenAI"):
provider = create_provider("cloudflare", settings)
assert isinstance(provider, CloudflareProvider)
diff --git a/uv.lock b/uv.lock
index 01823ba3..f9200921 100644
--- a/uv.lock
+++ b/uv.lock
@@ -350,34 +350,34 @@ wheels = [
[package.optional-dependencies]
cudart = [
- { name = "nvidia-cuda-runtime", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cuda-runtime" },
]
cufft = [
- { name = "nvidia-cufft", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cufft" },
]
cufile = [
- { name = "nvidia-cufile", marker = "sys_platform == 'linux'" },
+ { name = "nvidia-cufile" },
]
cupti = [
- { name = "nvidia-cuda-cupti", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cuda-cupti" },
]
curand = [
- { name = "nvidia-curand", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-curand" },
]
cusolver = [
- { name = "nvidia-cusolver", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cusolver" },
]
cusparse = [
- { name = "nvidia-cusparse", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cusparse" },
]
nvjitlink = [
- { name = "nvidia-nvjitlink", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-nvjitlink" },
]
nvrtc = [
- { name = "nvidia-cuda-nvrtc", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-cuda-nvrtc" },
]
nvtx = [
- { name = "nvidia-nvtx", marker = "sys_platform == 'linux' or sys_platform == 'win32'" },
+ { name = "nvidia-nvtx" },
]
[[package]]
@@ -561,7 +561,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "2.4.3"
+version = "2.4.4"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },