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

Add DeepInfra provider (#1396)

Ali Khokhar7f1c12b7af95
diff --git a/.env.example b/.env.example
index 8d83eec0..fd308270 100644
--- a/.env.example
+++ b/.env.example
@@ -117,6 +117,10 @@ QWENCLOUD_API_KEY=""
TOGETHER_API_KEY=""
+# DeepInfra (OpenAI-compatible Chat Completions at api.deepinfra.com/v1/openai)
+DEEPINFRA_API_KEY=""
+
+
# SambaNova Cloud (OpenAI-compatible Chat Completions at api.sambanova.ai/v1)
SAMBANOVA_API_KEY=""
@@ -147,7 +151,7 @@ OLLAMA_BASE_URL="http://localhost:11434"
# All Claude model requests are mapped to these models, plain model is fallback
# Format: provider_type/model/name
-# Valid providers: "nvidia_nim" | "openai" | "azure_openai" | "open_router" | "gemini" | "vertex" | "deepseek" | "mistral" | "mistral_codestral" | "opencode_zen" | "opencode_go" | "vercel" | "bedrock" | "huggingface" | "cohere" | "github_models" | "wafer" | "kimi" | "kimi_code" | "minimax" | "cerebras" | "groq" | "xai" | "qwencloud" | "together" | "sambanova" | "kilo" | "fireworks" | "novita" | "cloudflare" | "zai" | "ollama_cloud" | "lmstudio" | "llamacpp" | "ollama" | "tokenrouter" | "nararoute"
+# Valid providers: "nvidia_nim" | "openai" | "azure_openai" | "open_router" | "gemini" | "vertex" | "deepseek" | "mistral" | "mistral_codestral" | "opencode_zen" | "opencode_go" | "vercel" | "bedrock" | "huggingface" | "cohere" | "github_models" | "wafer" | "kimi" | "kimi_code" | "minimax" | "cerebras" | "groq" | "xai" | "qwencloud" | "together" | "deepinfra" | "sambanova" | "kilo" | "fireworks" | "novita" | "cloudflare" | "zai" | "ollama_cloud" | "lmstudio" | "llamacpp" | "ollama" | "tokenrouter" | "nararoute"
MODEL_FABLE=
MODEL_OPUS=
MODEL_SONNET=
@@ -191,6 +195,7 @@ FCC_SMOKE_MODEL_VERTEX=
FCC_SMOKE_MODEL_GROQ=
FCC_SMOKE_MODEL_QWENCLOUD=
FCC_SMOKE_MODEL_TOGETHER=
+FCC_SMOKE_MODEL_DEEPINFRA=
FCC_SMOKE_MODEL_SAMBANOVA=
FCC_SMOKE_MODEL_KILO=
FCC_SMOKE_MODEL_CEREBRAS=
@@ -217,6 +222,7 @@ OPENAI_PROXY=""
XAI_PROXY=""
QWENCLOUD_PROXY=""
TOGETHER_PROXY=""
+DEEPINFRA_PROXY=""
AZURE_OPENAI_PROXY=""
NVIDIA_NIM_PROXY=""
OPENROUTER_PROXY=""
diff --git a/README.md b/README.md
index 2c8191e0..6ee70f98 100644
--- a/README.md
+++ b/README.md
@@ -159,6 +159,7 @@ fcc-codex exec "hello"
| [xAI (Grok)](https://console.x.ai/team/default/api-keys) | `XAI_API_KEY` | `xai/grok-4.5` |
| [QwenCloud Token Plan](https://home.qwencloud.com/api-keys) | `QWENCLOUD_API_KEY` | `qwencloud/qwen3.7-plus` |
| [Together AI](https://api.together.ai/settings/api-keys) | `TOGETHER_API_KEY` | `together/zai-org/GLM-5.2` |
+| [DeepInfra](https://deepinfra.com/dash/api_keys) | `DEEPINFRA_API_KEY` | `deepinfra/deepseek-ai/DeepSeek-V4-Flash` |
| [Azure OpenAI](https://learn.microsoft.com/azure/foundry/openai/how-to/chatgpt) | `AZURE_OPENAI_API_KEY` and `AZURE_OPENAI_BASE_URL` | `azure_openai/<deployment-name>` |
| [Google AI Studio (Gemini)](https://aistudio.google.com/apikey) | `GEMINI_API_KEY` | `gemini/models/gemini-3.1-flash-lite` |
| [Google Vertex AI](https://cloud.google.com/vertex-ai/generative-ai/docs/start/openai) | `VERTEX_PROJECT_ID` + ADC | `vertex/google/gemini-3.5-flash` |
diff --git a/pyproject.toml b/pyproject.toml
index 3dbbc512..34866e1f 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "4.23.0"
+version = "4.24.0"
description = "Local proxy connecting coding agents to OpenAI-compatible AI providers"
readme = "README.md"
requires-python = ">=3.14.0"
diff --git a/smoke/lib/config.py b/smoke/lib/config.py
index e48aad37..0e162f25 100644
--- a/smoke/lib/config.py
+++ b/smoke/lib/config.py
@@ -73,6 +73,7 @@ PROVIDER_SMOKE_DEFAULT_MODELS: dict[str, str] = {
"xai": "xai/grok-4.5",
"qwencloud": "qwencloud/qwen3.7-plus",
"together": "together/zai-org/GLM-5.2",
+ "deepinfra": "deepinfra/deepseek-ai/DeepSeek-V4-Flash",
"sambanova": "sambanova/Meta-Llama-3.3-70B-Instruct",
"kilo": "kilo/kilo-auto/free",
"cerebras": "cerebras/llama3.1-8b",
diff --git a/src/free_claude_code/config/admin/provider_manifest.py b/src/free_claude_code/config/admin/provider_manifest.py
index 08e3b265..d296e4e9 100644
--- a/src/free_claude_code/config/admin/provider_manifest.py
+++ b/src/free_claude_code/config/admin/provider_manifest.py
@@ -159,6 +159,12 @@ _PROVIDER_FIELD_OVERRIDES: dict[str, dict[str, Any]] = {
"chat models."
),
},
+ "DEEPINFRA_API_KEY": {
+ "label": "DeepInfra API Key",
+ "description": (
+ "DeepInfra API key for OpenAI-compatible chat and reasoning models."
+ ),
+ },
"SAMBANOVA_API_KEY": {
"label": "SambaNova API Key",
"description": (
diff --git a/src/free_claude_code/config/provider_catalog.py b/src/free_claude_code/config/provider_catalog.py
index 9ef342cf..6158a250 100644
--- a/src/free_claude_code/config/provider_catalog.py
+++ b/src/free_claude_code/config/provider_catalog.py
@@ -58,6 +58,8 @@ QWENCLOUD_DEFAULT_BASE = (
)
# Together AI OpenAI-compatible Chat Completions API.
TOGETHER_DEFAULT_BASE = "https://api.together.ai/v1"
+# DeepInfra OpenAI-compatible Chat Completions API.
+DEEPINFRA_DEFAULT_BASE = "https://api.deepinfra.com/v1/openai"
# TokenRouter OpenAI-compatible Chat Completions gateway.
TOKENROUTER_DEFAULT_BASE = "https://api.tokenrouter.com/v1"
# NaraRoute OpenAI-compatible Chat Completions gateway.
@@ -161,6 +163,15 @@ PROVIDER_CATALOG: dict[str, ProviderDescriptor] = {
default_base_url=TOGETHER_DEFAULT_BASE,
proxy_attr="together_proxy",
),
+ "deepinfra": ProviderDescriptor(
+ provider_id="deepinfra",
+ display_name="DeepInfra",
+ credential_env="DEEPINFRA_API_KEY",
+ credential_url="https://deepinfra.com/dash/api_keys",
+ credential_attr="deepinfra_api_key",
+ default_base_url=DEEPINFRA_DEFAULT_BASE,
+ proxy_attr="deepinfra_proxy",
+ ),
"azure_openai": ProviderDescriptor(
provider_id="azure_openai",
display_name="Azure OpenAI",
diff --git a/src/free_claude_code/config/settings.py b/src/free_claude_code/config/settings.py
index bec89978..31be1520 100644
--- a/src/free_claude_code/config/settings.py
+++ b/src/free_claude_code/config/settings.py
@@ -140,6 +140,9 @@ class Settings(BaseSettings):
# ==================== Together AI (OpenAI-compatible) ====================
together_api_key: str = Field(default="", validation_alias="TOGETHER_API_KEY")
+ # ==================== DeepInfra (OpenAI-compatible) ====================
+ deepinfra_api_key: str = Field(default="", validation_alias="DEEPINFRA_API_KEY")
+
# ==================== Cerebras Inference (OpenAI-compatible) ====================
cerebras_api_key: str = Field(default="", validation_alias="CEREBRAS_API_KEY")
@@ -196,6 +199,7 @@ class Settings(BaseSettings):
xai_proxy: str = Field(default="", validation_alias="XAI_PROXY")
qwencloud_proxy: str = Field(default="", validation_alias="QWENCLOUD_PROXY")
together_proxy: str = Field(default="", validation_alias="TOGETHER_PROXY")
+ deepinfra_proxy: str = Field(default="", validation_alias="DEEPINFRA_PROXY")
azure_openai_proxy: str = Field(default="", validation_alias="AZURE_OPENAI_PROXY")
nvidia_nim_proxy: str = Field(default="", validation_alias="NVIDIA_NIM_PROXY")
open_router_proxy: str = Field(default="", validation_alias="OPENROUTER_PROXY")
diff --git a/src/free_claude_code/providers/model_listing.py b/src/free_claude_code/providers/model_listing.py
index 81860fcc..0620cdfd 100644
--- a/src/free_claude_code/providers/model_listing.py
+++ b/src/free_claude_code/providers/model_listing.py
@@ -32,23 +32,29 @@ def extract_openai_model_infos(
*,
provider_name: str,
collection_field: str | None = "data",
+ id_field: str = "id",
aliases_field: str | None = None,
field_equals: tuple[str, str] | None = None,
+ required_null_field: str | None = None,
+ tags_field: str | None = None,
+ thinking_tag: str = "reasoning",
+ non_thinking_tag: str | None = None,
) -> frozenset[_ProviderModelInfo]:
"""Extract routable IDs from an OpenAI-compatible model-list response."""
- model_ids: set[str] = set()
+ model_infos: set[_ProviderModelInfo] = set()
item_location = collection_field or "root-array"
for item in model_list_items(
payload,
provider_name=provider_name,
collection_field=collection_field,
):
- model_id = _field(item, "id")
+ model_id = _field(item, id_field)
if not isinstance(model_id, str) or not model_id.strip():
raise _malformed(
provider_name,
- f"expected every {item_location} item to include id",
+ f"expected every {item_location} item to include {id_field}",
)
+ included = True
if field_equals is not None:
field_name, expected_value = field_equals
field_value = _field(item, field_name)
@@ -59,8 +65,44 @@ def extract_openai_model_infos(
f"string {field_name}",
)
if field_value != expected_value:
- continue
- model_ids.add(model_id)
+ included = False
+
+ if required_null_field is not None:
+ if not _has_field(item, required_null_field):
+ raise _malformed(
+ provider_name,
+ f"expected every {item_location} item to include "
+ f"{required_null_field}",
+ )
+ if _field(item, required_null_field) is not None:
+ included = False
+
+ supports_thinking: bool | None = None
+ if tags_field is not None:
+ tags_value = _field(item, tags_field)
+ if not _is_sequence(tags_value) or any(
+ not isinstance(tag, str) or not tag.strip() for tag in tags_value
+ ):
+ raise _malformed(
+ provider_name,
+ f"expected every {item_location} item to include "
+ f"{tags_field} string array",
+ )
+ tags = frozenset(tags_value)
+ if thinking_tag in tags:
+ supports_thinking = True
+ elif non_thinking_tag is not None and non_thinking_tag in tags:
+ supports_thinking = False
+
+ if not included:
+ continue
+
+ model_infos.add(
+ _ProviderModelInfo(
+ model_id=model_id,
+ supports_thinking=supports_thinking,
+ )
+ )
if aliases_field is not None:
aliases = _field(item, aliases_field)
if not _is_sequence(aliases):
@@ -75,11 +117,16 @@ def extract_openai_model_infos(
provider_name,
f"expected every {aliases_field} item to be a model id",
)
- model_ids.add(alias)
+ model_infos.add(
+ _ProviderModelInfo(
+ model_id=alias,
+ supports_thinking=supports_thinking,
+ )
+ )
- if not model_ids:
+ if not model_infos:
raise _malformed(provider_name, "response did not include any model ids")
- return model_infos_from_ids(model_ids)
+ return frozenset(model_infos)
def extract_tool_capable_model_infos(
@@ -139,6 +186,12 @@ def _field(item: Any, name: str) -> Any:
return getattr(item, name, None)
+def _has_field(item: Any, name: str) -> bool:
+ if isinstance(item, Mapping):
+ return name in item
+ return hasattr(item, name)
+
+
def _is_sequence(value: Any) -> bool:
return isinstance(value, Sequence) and not isinstance(
value, str | bytes | bytearray
diff --git a/src/free_claude_code/providers/openai_chat/profiles.py b/src/free_claude_code/providers/openai_chat/profiles.py
index 3750bffe..da8df8de 100644
--- a/src/free_claude_code/providers/openai_chat/profiles.py
+++ b/src/free_claude_code/providers/openai_chat/profiles.py
@@ -60,8 +60,13 @@ class OpenAIModelListing:
path: str | None = None
collection_field: str | None = "data"
+ id_field: str = "id"
aliases_field: str | None = None
field_equals: tuple[str, str] | None = None
+ required_null_field: str | None = None
+ tags_field: str | None = None
+ thinking_tag: str = "reasoning"
+ non_thinking_tag: str | None = None
@dataclass(frozen=True, slots=True)
@@ -185,6 +190,30 @@ OPENAI_CHAT_PROFILES: dict[str, OpenAIChatProfile] = {
),
reasoning_delta_field="reasoning",
),
+ "deepinfra": OpenAIChatProfile(
+ _policy(
+ "DEEPINFRA",
+ ReasoningReplayMode.REASONING_CONTENT,
+ include_extra_body=True,
+ extra_body_validator=validate_extra_body_does_not_override_reasoning_fields,
+ default_max_tokens=ANTHROPIC_DEFAULT_MAX_OUTPUT_TOKENS,
+ ),
+ NamedEffortReasoning(
+ _ALL_EFFORTS,
+ disabled_value="none",
+ enabled_value="high",
+ use_extra_body=True,
+ ),
+ model_listing=OpenAIModelListing(
+ path="https://api.deepinfra.com/models/list",
+ collection_field=None,
+ id_field="model_name",
+ field_equals=("reported_type", "text-generation"),
+ required_null_field="deprecated",
+ tags_field="tags",
+ non_thinking_tag="non-reasoning",
+ ),
+ ),
"azure_openai": OpenAIChatProfile(
_policy(
"AZURE_OPENAI",
diff --git a/src/free_claude_code/providers/openai_chat/provider.py b/src/free_claude_code/providers/openai_chat/provider.py
index 8f824a28..aaf32d41 100644
--- a/src/free_claude_code/providers/openai_chat/provider.py
+++ b/src/free_claude_code/providers/openai_chat/provider.py
@@ -148,8 +148,13 @@ class OpenAIChatProvider(BaseProvider):
payload,
provider_name=self._provider_name,
collection_field=listing.collection_field,
+ id_field=listing.id_field,
aliases_field=listing.aliases_field,
field_equals=listing.field_equals,
+ required_null_field=listing.required_null_field,
+ tags_field=listing.tags_field,
+ thinking_tag=listing.thinking_tag,
+ non_thinking_tag=listing.non_thinking_tag,
)
async def _list_models_payload(self) -> Any:
diff --git a/tests/contracts/test_provider_catalog_order.py b/tests/contracts/test_provider_catalog_order.py
index 1f2bf280..7e6cfebb 100644
--- a/tests/contracts/test_provider_catalog_order.py
+++ b/tests/contracts/test_provider_catalog_order.py
@@ -13,6 +13,7 @@ _EXPECTED_PROVIDER_ORDER: tuple[str, ...] = (
"xai",
"qwencloud",
"together",
+ "deepinfra",
"azure_openai",
"gemini",
"vertex",
diff --git a/tests/contracts/test_smoke_config.py b/tests/contracts/test_smoke_config.py
index 1225c740..db4c8efb 100644
--- a/tests/contracts/test_smoke_config.py
+++ b/tests/contracts/test_smoke_config.py
@@ -56,6 +56,7 @@ def _settings(**overrides):
"xai_api_key": "",
"qwencloud_api_key": "",
"together_api_key": "",
+ "deepinfra_api_key": "",
"sambanova_api_key": "",
"cerebras_api_key": "",
"ollama_api_key": "",
@@ -205,6 +206,23 @@ def test_together_provider_smoke_uses_current_coding_model(monkeypatch) -> None:
assert models[0].source == "provider_default"
+def test_deepinfra_provider_smoke_uses_current_coding_model(monkeypatch) -> None:
+ monkeypatch.delenv("FCC_SMOKE_MODEL_DEEPINFRA", raising=False)
+ config = _smoke_config(
+ settings=_settings(
+ model="ollama/llama3.1",
+ ollama_base_url="",
+ deepinfra_api_key="deepinfra-key",
+ )
+ )
+
+ models = config.provider_smoke_models()
+
+ assert [model.provider for model in models] == ["deepinfra"]
+ assert models[0].full_model == "deepinfra/deepseek-ai/DeepSeek-V4-Flash"
+ assert models[0].source == "provider_default"
+
+
def test_connected_account_provider_smoke_requires_explicit_model(
monkeypatch,
) -> None:
diff --git a/tests/providers/test_deepinfra.py b/tests/providers/test_deepinfra.py
new file mode 100644
index 00000000..04550420
--- /dev/null
+++ b/tests/providers/test_deepinfra.py
@@ -0,0 +1,369 @@
+"""Tests for the DeepInfra OpenAI-chat profile and model catalog."""
+
+from types import SimpleNamespace
+from typing import Any
+from unittest.mock import AsyncMock, patch
+
+import httpx
+import pytest
+from openai import AsyncOpenAI
+
+from free_claude_code.application.errors import InvalidRequestError
+from free_claude_code.application.model_metadata import ProviderModelInfo
+from free_claude_code.config.constants import ANTHROPIC_DEFAULT_MAX_OUTPUT_TOKENS
+from free_claude_code.config.provider_catalog import DEEPINFRA_DEFAULT_BASE
+from free_claude_code.core.anthropic.models import MessagesRequest
+from free_claude_code.core.reasoning import ReasoningEffort, ReasoningPolicy
+from free_claude_code.providers.base import ProviderConfig
+from free_claude_code.providers.model_listing import ModelListResponseError
+from free_claude_code.providers.openai_chat import OpenAIChatProvider
+from tests.providers.support import (
+ REASONING_OFF,
+ REASONING_ON,
+ immediate_admission,
+ profiled_provider,
+ reasoning_for,
+)
+
+_CATALOG_URL = "https://api.deepinfra.com/models/list"
+_MODEL = "deepseek-ai/DeepSeek-V4-Flash"
+_MISSING = object()
+
+
+def _request(**overrides: Any) -> MessagesRequest:
+ payload: dict[str, Any] = {
+ "model": _MODEL,
+ "messages": [{"role": "user", "content": "Hello"}],
+ }
+ payload.update(overrides)
+ return MessagesRequest.model_validate(payload)
+
+
+def _catalog_model(
+ model_name: object = "model",
+ *,
+ reported_type: object = "text-generation",
+ deprecated: object = None,
+ tags: object = (),
+) -> dict[str, object]:
+ item: dict[str, object] = {}
+ for field, value in (
+ ("model_name", model_name),
+ ("reported_type", reported_type),
+ ("deprecated", deprecated),
+ ("tags", tags),
+ ):
+ if value is not _MISSING:
+ item[field] = (
+ list(value) if field == "tags" and isinstance(value, tuple) else value
+ )
+ return item
+
+
+def deepinfra_provider() -> OpenAIChatProvider:
+ return profiled_provider(
+ "deepinfra",
+ ProviderConfig(
+ api_key="test-deepinfra-key",
+ base_url=DEEPINFRA_DEFAULT_BASE,
+ rate_limit=10,
+ rate_window=60,
+ ),
+ admission=immediate_admission(provider_name="deepinfra"),
+ )
+
+
+def test_init_uses_openai_chat_provider(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ assert deepinfra_provider._api_key == "test-deepinfra-key"
+ assert deepinfra_provider._base_url == DEEPINFRA_DEFAULT_BASE
+ assert deepinfra_provider._provider_name == "DEEPINFRA"
+
+
+def test_build_request_body_preserves_common_chat_tools_and_images(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ request = MessagesRequest.model_validate(
+ {
+ "model": "deepseek-ai/DeepSeek-V4-Flash",
+ "system": "Be concise.",
+ "messages": [
+ {
+ "role": "user",
+ "content": [
+ {"type": "text", "text": "Inspect this image."},
+ {
+ "type": "image",
+ "source": {
+ "type": "base64",
+ "media_type": "image/png",
+ "data": "aW1hZ2U=",
+ },
+ },
+ ],
+ }
+ ],
+ "tools": [
+ {
+ "name": "inspect_image",
+ "description": "Inspect an image",
+ "input_schema": {
+ "type": "object",
+ "properties": {"detail": {"type": "string"}},
+ },
+ }
+ ],
+ }
+ )
+
+ body = deepinfra_provider._build_request_body(
+ request,
+ reasoning=reasoning_for(request),
+ )
+
+ assert body["model"] == "deepseek-ai/DeepSeek-V4-Flash"
+ assert body["max_tokens"] == ANTHROPIC_DEFAULT_MAX_OUTPUT_TOKENS
+ assert body["messages"][0] == {"role": "system", "content": "Be concise."}
+ assert body["messages"][1]["content"] == [
+ {"type": "text", "text": "Inspect this image."},
+ {
+ "type": "image_url",
+ "image_url": {"url": "data:image/png;base64,aW1hZ2U="},
+ },
+ ]
+ assert body["tools"][0]["function"]["name"] == "inspect_image"
+
+
+ ("reasoning", "expected"),
+ [
+ (REASONING_OFF, "none"),
+ (REASONING_ON, "high"),
+ *[
+ (ReasoningPolicy.on(effort=effort), effort.value)
+ for effort in ReasoningEffort
+ ],
+ ],
+)
+def test_build_request_body_encodes_documented_reasoning_effort(
+ deepinfra_provider: OpenAIChatProvider,
+ reasoning: ReasoningPolicy,
+ expected: str,
+) -> None:
+ request = _request()
+
+ body = deepinfra_provider._build_request_body(request, reasoning=reasoning)
+
+ assert body["extra_body"] == {"reasoning_effort": expected}
+
+
+def test_build_request_body_preserves_extra_body_without_reasoning_override(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ request = _request(extra_body={"service_tier": "priority"})
+
+ body = deepinfra_provider._build_request_body(request, reasoning=REASONING_ON)
+
+ assert body["extra_body"] == {
+ "service_tier": "priority",
+ "reasoning_effort": "high",
+ }
+
+
[email protected]("field", ["reasoning", "reasoning_effort"])
+def test_build_request_body_rejects_caller_reasoning_override(
+ deepinfra_provider: OpenAIChatProvider,
+ field: str,
+) -> None:
+ request = _request(extra_body={field: "caller-owned"})
+
+ with pytest.raises(InvalidRequestError, match="must not override reasoning"):
+ deepinfra_provider._build_request_body(request, reasoning=REASONING_ON)
+
+
+def test_build_request_body_replays_documented_reasoning_content(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ request = MessagesRequest.model_validate(
+ {
+ "model": "deepseek-ai/DeepSeek-V4-Flash",
+ "messages": [
+ {"role": "user", "content": "Solve it."},
+ {
+ "role": "assistant",
+ "content": [
+ {"type": "thinking", "thinking": "Work through it."},
+ {"type": "text", "text": "The answer is 42."},
+ ],
+ },
+ {"role": "user", "content": "Continue."},
+ ],
+ }
+ )
+
+ body = deepinfra_provider._build_request_body(
+ request,
+ reasoning=reasoning_for(request),
+ )
+
+ assert body["messages"][1] == {
+ "role": "assistant",
+ "content": "The answer is 42.",
+ "reasoning_content": "Work through it.",
+ }
+ assert (
+ deepinfra_provider._profile.reasoning_delta(
+ SimpleNamespace(reasoning_content="next thought")
+ )
+ == "next thought"
+ )
+
+
+async def test_lists_only_active_text_models_with_thinking_metadata(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ deepinfra_provider._client.get = AsyncMock(
+ return_value=[
+ _catalog_model("reasoning-model", tags=("openai", "reasoning", "tools")),
+ _catalog_model("plain-model", tags=("openai", "non-reasoning", "tools")),
+ _catalog_model("hybrid-model", tags=("reasoning", "non-reasoning")),
+ _catalog_model("unknown-model", tags=("openai",)),
+ _catalog_model(
+ "deprecated-model",
+ deprecated=1_700_000_000,
+ tags=("reasoning",),
+ ),
+ _catalog_model("image-model", reported_type="text-to-image"),
+ ]
+ )
+
+ model_infos = await deepinfra_provider.list_model_infos()
+
+ assert model_infos == frozenset(
+ {
+ ProviderModelInfo("reasoning-model", supports_thinking=True),
+ ProviderModelInfo("plain-model", supports_thinking=False),
+ ProviderModelInfo("hybrid-model", supports_thinking=True),
+ ProviderModelInfo("unknown-model"),
+ }
+ )
+ deepinfra_provider._client.get.assert_awaited_once_with(
+ _CATALOG_URL,
+ cast_to=object,
+ )
+
+
+async def test_model_catalog_uses_absolute_public_url() -> None:
+ requests: list[httpx.Request] = []
+
+ def handler(request: httpx.Request) -> httpx.Response:
+ requests.append(request)
+ return httpx.Response(
+ 200,
+ json=[_catalog_model(_MODEL, tags=("reasoning", "tools"))],
+ )
+
+ http_client = httpx.AsyncClient(transport=httpx.MockTransport(handler))
+
+ def build_client(*args: Any, **kwargs: Any) -> AsyncOpenAI:
+ kwargs["http_client"] = http_client
+ return AsyncOpenAI(*args, **kwargs)
+
+ with patch(
+ "free_claude_code.providers.openai_chat.provider.AsyncOpenAI",
+ side_effect=build_client,
+ ):
+ provider = profiled_provider(
+ "deepinfra",
+ ProviderConfig(
+ api_key="wire-deepinfra-key",
+ base_url=DEEPINFRA_DEFAULT_BASE,
+ rate_limit=10,
+ rate_window=60,
+ ),
+ admission=immediate_admission(provider_name="deepinfra"),
+ )
+ try:
+ model_infos = await provider.list_model_infos()
+ finally:
+ await provider.cleanup()
+
+ assert model_infos == frozenset(
+ {
+ ProviderModelInfo(
+ "deepseek-ai/DeepSeek-V4-Flash",
+ supports_thinking=True,
+ )
+ }
+ )
+ assert len(requests) == 1
+ assert str(requests[0].url) == _CATALOG_URL
+ assert requests[0].headers["authorization"] == "Bearer wire-deepinfra-key"
+
+
+ ("payload", "message"),
+ [
+ ({"data": []}, "expected root array"),
+ ([_catalog_model(model_name=_MISSING)], "include model_name"),
+ ([_catalog_model(reported_type=_MISSING)], "include string reported_type"),
+ ([_catalog_model(reported_type=7)], "include string reported_type"),
+ ([_catalog_model(deprecated=_MISSING)], "include deprecated"),
+ ([_catalog_model(tags=_MISSING)], "include tags string array"),
+ ([_catalog_model(tags=[7])], "include tags string array"),
+ ([], "did not include any model ids"),
+ (
+ [_catalog_model("image", reported_type="text-to-image")],
+ "did not include any model ids",
+ ),
+ (
+ [_catalog_model("old", deprecated=1)],
+ "did not include any model ids",
+ ),
+ ],
+)
+async def test_rejects_malformed_or_unusable_catalog_atomically(
+ deepinfra_provider: OpenAIChatProvider,
+ payload: object,
+ message: str,
+) -> None:
+ deepinfra_provider._client.get = AsyncMock(return_value=payload)
+
+ with pytest.raises(ModelListResponseError, match=message):
+ await deepinfra_provider.list_model_infos()
+
+
+async def test_model_catalog_uses_shared_retry_admission(
+ deepinfra_provider: OpenAIChatProvider,
+) -> None:
+ request = httpx.Request("GET", _CATALOG_URL)
+ response = httpx.Response(429, request=request)
+ rate_limit_error = httpx.HTTPStatusError(
+ "rate limited",
+ request=request,
+ response=response,
+ )
+ deepinfra_provider._client.get = AsyncMock(
+ side_effect=[
+ rate_limit_error,
+ [_catalog_model(_MODEL, tags=("reasoning", "tools"))],
+ ]
+ )
+
+ model_infos = await deepinfra_provider.list_model_infos()
+
+ assert model_infos == frozenset(
+ {
+ ProviderModelInfo(
+ "deepseek-ai/DeepSeek-V4-Flash",
+ supports_thinking=True,
+ )
+ }
+ )
+ assert deepinfra_provider._client.get.await_count == 2
diff --git a/tests/providers/test_provider_runtime.py b/tests/providers/test_provider_runtime.py
index d34f08d9..6250e0c3 100644
--- a/tests/providers/test_provider_runtime.py
+++ b/tests/providers/test_provider_runtime.py
@@ -13,6 +13,7 @@ from free_claude_code.config.nim import NimSettings
from free_claude_code.config.provider_catalog import (
BEDROCK_DEFAULT_BASE,
COHERE_DEFAULT_BASE,
+ DEEPINFRA_DEFAULT_BASE,
GITHUB_MODELS_DEFAULT_BASE,
HUGGINGFACE_DEFAULT_BASE,
KIMI_CODE_DEFAULT_BASE,
@@ -66,6 +67,7 @@ def _make_settings(**overrides):
mock.xai_api_key = "test_xai_key"
mock.qwencloud_api_key = "test_qwencloud_key"
mock.together_api_key = "test_together_key"
+ mock.deepinfra_api_key = "test_deepinfra_key"
mock.mistral_api_key = "test_mistral_key"
mock.codestral_api_key = "test_codestral_key"
mock.deepseek_api_key = "test_deepseek_key"
@@ -132,6 +134,7 @@ def _make_settings(**overrides):
mock.xai_proxy = ""
mock.qwencloud_proxy = ""
mock.together_proxy = ""
+ mock.deepinfra_proxy = ""
mock.azure_openai_proxy = ""
mock.provider_rate_limit = 40
mock.provider_rate_window = 60
@@ -258,6 +261,25 @@ def test_together_provider_config_uses_key_base_and_proxy() -> None:
assert isinstance(provider, OpenAIChatProvider)
+def test_deepinfra_provider_config_uses_key_base_and_proxy() -> None:
+ descriptor = PROVIDER_CATALOG["deepinfra"]
+ settings = _make_settings(
+ deepinfra_api_key="deepinfra-token",
+ deepinfra_proxy="http://proxy.test:8080",
+ )
+
+ config = build_provider_config(descriptor, settings)
+ with patch("free_claude_code.providers.openai_chat.provider.AsyncOpenAI"):
+ provider = create_provider("deepinfra", settings)
+
+ assert descriptor.display_name == "DeepInfra"
+ assert descriptor.credential_env == "DEEPINFRA_API_KEY"
+ assert config.api_key == "deepinfra-token"
+ assert config.base_url == DEEPINFRA_DEFAULT_BASE
+ assert config.proxy == "http://proxy.test:8080"
+ assert isinstance(provider, OpenAIChatProvider)
+
+
def test_bedrock_provider_config_uses_regional_base_key_and_proxy() -> None:
descriptor = PROVIDER_CATALOG["bedrock"]
settings = _make_settings(
@@ -552,6 +574,7 @@ def test_create_provider_instantiates_each_builtin():
"xai": OpenAIChatProvider,
"qwencloud": OpenAIChatProvider,
"together": OpenAIChatProvider,
+ "deepinfra": OpenAIChatProvider,
"azure_openai": OpenAIChatProvider,
"open_router": OpenRouterProvider,
"mistral": MistralProvider,
diff --git a/uv.lock b/uv.lock
index b5d67bf5..1043072f 100644
--- a/uv.lock
+++ b/uv.lock
@@ -620,7 +620,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "4.23.0"
+version = "4.24.0"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },