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 Together AI provider (#1395)

Ali Khokhar5586754de35a
diff --git a/.env.example b/.env.example
index bb68fb59..8d83eec0 100644
--- a/.env.example
+++ b/.env.example
@@ -113,6 +113,10 @@ XAI_API_KEY=""
QWENCLOUD_API_KEY=""
+# Together AI (OpenAI-compatible Chat Completions at api.together.ai/v1)
+TOGETHER_API_KEY=""
+
+
# SambaNova Cloud (OpenAI-compatible Chat Completions at api.sambanova.ai/v1)
SAMBANOVA_API_KEY=""
@@ -143,7 +147,7 @@ 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" | "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" | "sambanova" | "kilo" | "fireworks" | "novita" | "cloudflare" | "zai" | "ollama_cloud" | "lmstudio" | "llamacpp" | "ollama" | "tokenrouter" | "nararoute"
MODEL_FABLE=
MODEL_OPUS=
MODEL_SONNET=
@@ -186,6 +190,7 @@ FCC_SMOKE_MODEL_GEMINI=
FCC_SMOKE_MODEL_VERTEX=
FCC_SMOKE_MODEL_GROQ=
FCC_SMOKE_MODEL_QWENCLOUD=
+FCC_SMOKE_MODEL_TOGETHER=
FCC_SMOKE_MODEL_SAMBANOVA=
FCC_SMOKE_MODEL_KILO=
FCC_SMOKE_MODEL_CEREBRAS=
@@ -211,6 +216,7 @@ REASONING_HAIKU=inherit
OPENAI_PROXY=""
XAI_PROXY=""
QWENCLOUD_PROXY=""
+TOGETHER_PROXY=""
AZURE_OPENAI_PROXY=""
NVIDIA_NIM_PROXY=""
OPENROUTER_PROXY=""
diff --git a/README.md b/README.md
index 9bfebb90..2c8191e0 100644
--- a/README.md
+++ b/README.md
@@ -158,6 +158,7 @@ fcc-codex exec "hello"
| [OpenAI / ChatGPT](https://learn.chatgpt.com/docs/auth) | Connect ChatGPT in the Admin UI | `openai/<model-id>` |
| [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` |
| [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 4ed4f4de..3dbbc512 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "4.22.0"
+version = "4.23.0"
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 c810494f..e48aad37 100644
--- a/smoke/lib/config.py
+++ b/smoke/lib/config.py
@@ -72,6 +72,7 @@ PROVIDER_SMOKE_DEFAULT_MODELS: dict[str, str] = {
"groq": "groq/llama-3.3-70b-versatile",
"xai": "xai/grok-4.5",
"qwencloud": "qwencloud/qwen3.7-plus",
+ "together": "together/zai-org/GLM-5.2",
"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 7eb05181..08e3b265 100644
--- a/src/free_claude_code/config/admin/provider_manifest.py
+++ b/src/free_claude_code/config/admin/provider_manifest.py
@@ -152,6 +152,13 @@ _PROVIDER_FIELD_OVERRIDES: dict[str, dict[str, Any]] = {
"mixed."
),
},
+ "TOGETHER_API_KEY": {
+ "label": "Together AI API Key",
+ "description": (
+ "Together AI OpenAI-compatible API key for serverless and dedicated "
+ "chat 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 239ad0d2..9ef342cf 100644
--- a/src/free_claude_code/config/provider_catalog.py
+++ b/src/free_claude_code/config/provider_catalog.py
@@ -56,6 +56,8 @@ XAI_DEFAULT_BASE = "https://api.x.ai/v1"
QWENCLOUD_DEFAULT_BASE = (
"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"
)
+# Together AI OpenAI-compatible Chat Completions API.
+TOGETHER_DEFAULT_BASE = "https://api.together.ai/v1"
# TokenRouter OpenAI-compatible Chat Completions gateway.
TOKENROUTER_DEFAULT_BASE = "https://api.tokenrouter.com/v1"
# NaraRoute OpenAI-compatible Chat Completions gateway.
@@ -150,6 +152,15 @@ PROVIDER_CATALOG: dict[str, ProviderDescriptor] = {
default_base_url=QWENCLOUD_DEFAULT_BASE,
proxy_attr="qwencloud_proxy",
),
+ "together": ProviderDescriptor(
+ provider_id="together",
+ display_name="Together AI",
+ credential_env="TOGETHER_API_KEY",
+ credential_url="https://api.together.ai/settings/api-keys",
+ credential_attr="together_api_key",
+ default_base_url=TOGETHER_DEFAULT_BASE,
+ proxy_attr="together_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 cef0e03a..bec89978 100644
--- a/src/free_claude_code/config/settings.py
+++ b/src/free_claude_code/config/settings.py
@@ -137,6 +137,9 @@ class Settings(BaseSettings):
# ==================== QwenCloud Token Plan (OpenAI-compatible) ====================
qwencloud_api_key: str = Field(default="", validation_alias="QWENCLOUD_API_KEY")
+ # ==================== Together AI (OpenAI-compatible) ====================
+ together_api_key: str = Field(default="", validation_alias="TOGETHER_API_KEY")
+
# ==================== Cerebras Inference (OpenAI-compatible) ====================
cerebras_api_key: str = Field(default="", validation_alias="CEREBRAS_API_KEY")
@@ -192,6 +195,7 @@ class Settings(BaseSettings):
openai_proxy: str = Field(default="", validation_alias="OPENAI_PROXY")
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")
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 0cb0d68e..81860fcc 100644
--- a/src/free_claude_code/providers/model_listing.py
+++ b/src/free_claude_code/providers/model_listing.py
@@ -31,11 +31,13 @@ def extract_openai_model_infos(
payload: Any,
*,
provider_name: str,
- collection_field: str = "data",
+ collection_field: str | None = "data",
aliases_field: str | None = None,
+ field_equals: tuple[str, str] | None = None,
) -> frozenset[_ProviderModelInfo]:
"""Extract routable IDs from an OpenAI-compatible model-list response."""
model_ids: set[str] = set()
+ item_location = collection_field or "root-array"
for item in model_list_items(
payload,
provider_name=provider_name,
@@ -45,15 +47,26 @@ def extract_openai_model_infos(
if not isinstance(model_id, str) or not model_id.strip():
raise _malformed(
provider_name,
- f"expected every {collection_field} item to include id",
+ f"expected every {item_location} item to include id",
)
+ if field_equals is not None:
+ field_name, expected_value = field_equals
+ field_value = _field(item, field_name)
+ if not isinstance(field_value, str):
+ raise _malformed(
+ provider_name,
+ f"expected every {item_location} item to include "
+ f"string {field_name}",
+ )
+ if field_value != expected_value:
+ continue
model_ids.add(model_id)
if aliases_field is not None:
aliases = _field(item, aliases_field)
if not _is_sequence(aliases):
raise _malformed(
provider_name,
- f"expected every {collection_field} item to include "
+ f"expected every {item_location} item to include "
f"{aliases_field} array",
)
for alias in aliases:
@@ -103,14 +116,19 @@ def model_list_items(
payload: Any,
*,
provider_name: str,
- collection_field: str = "data",
+ collection_field: str | None = "data",
) -> tuple[Any, ...]:
"""Return a validated OpenAI-shaped model-list data array."""
- data = _field(payload, collection_field)
+ data = payload if collection_field is None else _field(payload, collection_field)
if not _is_sequence(data):
+ location = (
+ "root array"
+ if collection_field is None
+ else (f"top-level {collection_field} array")
+ )
raise _malformed(
provider_name,
- f"expected top-level {collection_field} array",
+ f"expected {location}",
)
return tuple(data)
diff --git a/src/free_claude_code/providers/openai_chat/profiles.py b/src/free_claude_code/providers/openai_chat/profiles.py
index 2bc80d66..3750bffe 100644
--- a/src/free_claude_code/providers/openai_chat/profiles.py
+++ b/src/free_claude_code/providers/openai_chat/profiles.py
@@ -59,8 +59,9 @@ class OpenAIModelListing:
"""Declarative model-list endpoint and response shape."""
path: str | None = None
- collection_field: str = "data"
+ collection_field: str | None = "data"
aliases_field: str | None = None
+ field_equals: tuple[str, str] | None = None
@dataclass(frozen=True, slots=True)
@@ -170,6 +171,20 @@ OPENAI_CHAT_PROFILES: dict[str, OpenAIChatProfile] = {
),
NO_REASONING,
),
+ "together": OpenAIChatProfile(
+ _policy(
+ "TOGETHER",
+ ReasoningReplayMode.REASONING,
+ default_max_tokens=ANTHROPIC_DEFAULT_MAX_OUTPUT_TOKENS,
+ ),
+ NO_REASONING,
+ model_listing=OpenAIModelListing(
+ path="/models",
+ collection_field=None,
+ field_equals=("type", "chat"),
+ ),
+ reasoning_delta_field="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 5d7bce0f..8f824a28 100644
--- a/src/free_claude_code/providers/openai_chat/provider.py
+++ b/src/free_claude_code/providers/openai_chat/provider.py
@@ -149,6 +149,7 @@ class OpenAIChatProvider(BaseProvider):
provider_name=self._provider_name,
collection_field=listing.collection_field,
aliases_field=listing.aliases_field,
+ field_equals=listing.field_equals,
)
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 7e0353fa..1f2bf280 100644
--- a/tests/contracts/test_provider_catalog_order.py
+++ b/tests/contracts/test_provider_catalog_order.py
@@ -12,6 +12,7 @@ _EXPECTED_PROVIDER_ORDER: tuple[str, ...] = (
"openai",
"xai",
"qwencloud",
+ "together",
"azure_openai",
"gemini",
"vertex",
diff --git a/tests/contracts/test_smoke_config.py b/tests/contracts/test_smoke_config.py
index 860e5de6..1225c740 100644
--- a/tests/contracts/test_smoke_config.py
+++ b/tests/contracts/test_smoke_config.py
@@ -55,6 +55,7 @@ def _settings(**overrides):
"groq_api_key": "",
"xai_api_key": "",
"qwencloud_api_key": "",
+ "together_api_key": "",
"sambanova_api_key": "",
"cerebras_api_key": "",
"ollama_api_key": "",
@@ -187,6 +188,23 @@ def test_qwencloud_provider_smoke_uses_current_coding_model(monkeypatch) -> None
assert models[0].source == "provider_default"
+def test_together_provider_smoke_uses_current_coding_model(monkeypatch) -> None:
+ monkeypatch.delenv("FCC_SMOKE_MODEL_TOGETHER", raising=False)
+ config = _smoke_config(
+ settings=_settings(
+ model="ollama/llama3.1",
+ ollama_base_url="",
+ together_api_key="together-key",
+ )
+ )
+
+ models = config.provider_smoke_models()
+
+ assert [model.provider for model in models] == ["together"]
+ assert models[0].full_model == "together/zai-org/GLM-5.2"
+ assert models[0].source == "provider_default"
+
+
def test_connected_account_provider_smoke_requires_explicit_model(
monkeypatch,
) -> None:
diff --git a/tests/providers/test_provider_runtime.py b/tests/providers/test_provider_runtime.py
index 92d9900d..d34f08d9 100644
--- a/tests/providers/test_provider_runtime.py
+++ b/tests/providers/test_provider_runtime.py
@@ -22,6 +22,7 @@ from free_claude_code.config.provider_catalog import (
PROVIDER_CATALOG,
QWENCLOUD_DEFAULT_BASE,
SUPPORTED_PROVIDER_IDS,
+ TOGETHER_DEFAULT_BASE,
TOKENROUTER_DEFAULT_BASE,
VERCEL_AI_GATEWAY_DEFAULT_BASE,
XAI_DEFAULT_BASE,
@@ -64,6 +65,7 @@ def _make_settings(**overrides):
mock.open_router_api_key = "test_openrouter_key"
mock.xai_api_key = "test_xai_key"
mock.qwencloud_api_key = "test_qwencloud_key"
+ mock.together_api_key = "test_together_key"
mock.mistral_api_key = "test_mistral_key"
mock.codestral_api_key = "test_codestral_key"
mock.deepseek_api_key = "test_deepseek_key"
@@ -129,6 +131,7 @@ def _make_settings(**overrides):
mock.openai_proxy = ""
mock.xai_proxy = ""
mock.qwencloud_proxy = ""
+ mock.together_proxy = ""
mock.azure_openai_proxy = ""
mock.provider_rate_limit = 40
mock.provider_rate_window = 60
@@ -236,6 +239,25 @@ def test_qwencloud_provider_config_uses_key_base_and_proxy() -> None:
assert isinstance(provider, OpenAIChatProvider)
+def test_together_provider_config_uses_key_base_and_proxy() -> None:
+ descriptor = PROVIDER_CATALOG["together"]
+ settings = _make_settings(
+ together_api_key="together-token",
+ together_proxy="http://proxy.test:8080",
+ )
+
+ config = build_provider_config(descriptor, settings)
+ with patch("free_claude_code.providers.openai_chat.provider.AsyncOpenAI"):
+ provider = create_provider("together", settings)
+
+ assert descriptor.display_name == "Together AI"
+ assert descriptor.credential_env == "TOGETHER_API_KEY"
+ assert config.api_key == "together-token"
+ assert config.base_url == TOGETHER_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(
@@ -529,6 +551,7 @@ def test_create_provider_instantiates_each_builtin():
"openai": OpenAICodexProvider,
"xai": OpenAIChatProvider,
"qwencloud": OpenAIChatProvider,
+ "together": OpenAIChatProvider,
"azure_openai": OpenAIChatProvider,
"open_router": OpenRouterProvider,
"mistral": MistralProvider,
diff --git a/tests/providers/test_together.py b/tests/providers/test_together.py
new file mode 100644
index 00000000..605a2623
--- /dev/null
+++ b/tests/providers/test_together.py
@@ -0,0 +1,277 @@
+"""Tests for the Together AI OpenAI-chat profile and model catalog."""
+
+from types import SimpleNamespace
+from unittest.mock import AsyncMock
+
+import httpx
+import pytest
+from openai import AsyncOpenAI
+
+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 TOGETHER_DEFAULT_BASE
+from free_claude_code.core.anthropic.models import MessagesRequest
+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,
+)
+
+
+def together_provider() -> OpenAIChatProvider:
+ return profiled_provider(
+ "together",
+ ProviderConfig(
+ api_key="test-together-key",
+ base_url=TOGETHER_DEFAULT_BASE,
+ rate_limit=10,
+ rate_window=60,
+ ),
+ admission=immediate_admission(provider_name="together"),
+ )
+
+
+def test_init_uses_openai_chat_provider(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ assert together_provider._api_key == "test-together-key"
+ assert together_provider._base_url == TOGETHER_DEFAULT_BASE
+ assert together_provider._provider_name == "TOGETHER"
+
+
+def test_build_request_body_preserves_common_chat_tools_and_images(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ request = MessagesRequest.model_validate(
+ {
+ "model": "zai-org/GLM-5.2",
+ "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 = together_provider._build_request_body(
+ request,
+ reasoning=reasoning_for(request),
+ )
+
+ assert body["model"] == "zai-org/GLM-5.2"
+ 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"
+
+
[email protected]("reasoning", [REASONING_OFF, REASONING_ON])
+def test_build_request_body_does_not_invent_catalog_wide_reasoning_control(
+ together_provider: OpenAIChatProvider,
+ reasoning,
+) -> None:
+ request = MessagesRequest.model_validate(
+ {
+ "model": "zai-org/GLM-5.2",
+ "messages": [{"role": "user", "content": "Hello"}],
+ }
+ )
+
+ body = together_provider._build_request_body(request, reasoning=reasoning)
+ extra_body = body.get("extra_body", {})
+
+ for field in ("reasoning", "reasoning_effort", "chat_template_kwargs"):
+ assert field not in body
+ assert field not in extra_body
+
+
+def test_build_request_body_replays_documented_reasoning_field(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ request = MessagesRequest.model_validate(
+ {
+ "model": "zai-org/GLM-5.2",
+ "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 = together_provider._build_request_body(
+ request,
+ reasoning=reasoning_for(request),
+ )
+
+ assert body["messages"][1] == {
+ "role": "assistant",
+ "content": "The answer is 42.",
+ "reasoning": "Work through it.",
+ }
+ assert (
+ together_provider._profile.reasoning_delta(
+ SimpleNamespace(reasoning="next thought")
+ )
+ == "next thought"
+ )
+
+
+async def test_lists_only_documented_chat_models(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ together_provider._client.get = AsyncMock(
+ return_value=[
+ {"id": "zai-org/GLM-5.2", "type": "chat"},
+ {"id": "openai/gpt-oss-120b", "type": "chat"},
+ {"id": "codellama/CodeLlama-70b", "type": "code"},
+ {"id": "black-forest-labs/FLUX.2-dev", "type": "image"},
+ {"id": "intfloat/multilingual-e5-large", "type": "embedding"},
+ {"id": "meta-llama/Llama-Guard-4-12B", "type": "moderation"},
+ {"id": "mixedbread-ai/Mxbai-Rerank-Large-V2", "type": "rerank"},
+ ]
+ )
+ together_provider._client.models.list = AsyncMock()
+
+ model_infos = await together_provider.list_model_infos()
+
+ assert model_infos == frozenset(
+ {
+ ProviderModelInfo("zai-org/GLM-5.2"),
+ ProviderModelInfo("openai/gpt-oss-120b"),
+ }
+ )
+ together_provider._client.get.assert_awaited_once_with(
+ "/models",
+ cast_to=object,
+ )
+ together_provider._client.models.list.assert_not_awaited()
+
+
+async def test_model_catalog_uses_configured_base_url_and_auth(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ requests: list[httpx.Request] = []
+
+ def handler(request: httpx.Request) -> httpx.Response:
+ requests.append(request)
+ return httpx.Response(
+ 200,
+ json=[
+ {
+ "id": "zai-org/GLM-5.2",
+ "object": "model",
+ "created": 0,
+ "type": "chat",
+ }
+ ],
+ )
+
+ await together_provider._client.close()
+ together_provider._client = AsyncOpenAI(
+ api_key="wire-together-key",
+ base_url=TOGETHER_DEFAULT_BASE,
+ max_retries=0,
+ http_client=httpx.AsyncClient(transport=httpx.MockTransport(handler)),
+ )
+ try:
+ model_infos = await together_provider.list_model_infos()
+ finally:
+ await together_provider.cleanup()
+
+ assert model_infos == frozenset({ProviderModelInfo("zai-org/GLM-5.2")})
+ assert len(requests) == 1
+ assert str(requests[0].url) == "https://api.together.ai/v1/models"
+ assert requests[0].headers["authorization"] == "Bearer wire-together-key"
+
+
+ ("payload", "message"),
+ [
+ ({"data": []}, "expected root array"),
+ ([{"type": "chat"}], "include id"),
+ ([{"id": "model"}], "include string type"),
+ ([{"id": "model", "type": 7}], "include string type"),
+ ([], "did not include any model ids"),
+ (
+ [{"id": "black-forest-labs/FLUX.2-dev", "type": "image"}],
+ "did not include any model ids",
+ ),
+ ],
+)
+async def test_rejects_malformed_or_unusable_catalog_atomically(
+ together_provider: OpenAIChatProvider,
+ payload: object,
+ message: str,
+) -> None:
+ together_provider._client.get = AsyncMock(return_value=payload)
+
+ with pytest.raises(ModelListResponseError, match=message):
+ await together_provider.list_model_infos()
+
+
+async def test_model_catalog_uses_shared_retry_admission(
+ together_provider: OpenAIChatProvider,
+) -> None:
+ request = httpx.Request("GET", "https://api.together.ai/v1/models")
+ response = httpx.Response(429, request=request)
+ rate_limit_error = httpx.HTTPStatusError(
+ "rate limited",
+ request=request,
+ response=response,
+ )
+ together_provider._client.get = AsyncMock(
+ side_effect=[
+ rate_limit_error,
+ [{"id": "zai-org/GLM-5.2", "type": "chat"}],
+ ]
+ )
+
+ model_infos = await together_provider.list_model_infos()
+
+ assert model_infos == frozenset({ProviderModelInfo("zai-org/GLM-5.2")})
+ assert together_provider._client.get.await_count == 2
diff --git a/uv.lock b/uv.lock
index d7801129..b5d67bf5 100644
--- a/uv.lock
+++ b/uv.lock
@@ -620,7 +620,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "4.22.0"
+version = "4.23.0"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },