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: restore reliable live smoke coverage (#1608)

Ali Khokhar1f002bb7571a
diff --git a/pyproject.toml b/pyproject.toml
index d584a33f..14605805 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "5.17.0"
+version = "5.17.1"
description = "Local proxy connecting coding agents to OpenAI-compatible AI providers"
readme = "README.md"
requires-python = ">=3.14.0"
diff --git a/smoke/README.md b/smoke/README.md
index c232878f..41c13322 100644
--- a/smoke/README.md
+++ b/smoke/README.md
@@ -103,14 +103,14 @@ uv run pytest smoke/product -n 0 -s --tb=short
```powershell
$env:FCC_LIVE_SMOKE = "1"
$env:FCC_SMOKE_TARGETS = "nvidia_nim_cli"
-$env:FCC_SMOKE_NIM_MODELS = "z-ai/glm-5.2,moonshotai/kimi-k2.6,minimaxai/minimax-m2.7,minimaxai/minimax-m3,nvidia/nemotron-3-super-120b-a12b,deepseek-ai/deepseek-v4-pro,deepseek-ai/deepseek-v4-flash"
+$env:FCC_SMOKE_NIM_MODELS = "nvidia/nemotron-3.5-lightning-30b-a3b,moonshotai/kimi-k3,minimaxai/minimax-m3,nvidia/nemotron-3-super-120b-a12b"
uv run pytest smoke/product -n 0 -s --tb=short
```
```powershell
$env:FCC_LIVE_SMOKE = "1"
$env:FCC_SMOKE_TARGETS = "openrouter_free_cli"
-$env:FCC_SMOKE_OPENROUTER_FREE_MODELS = "nvidia/nemotron-3-super-120b-a12b:free,openai/gpt-oss-120b:free,poolside/laguna-m.1:free"
+$env:FCC_SMOKE_OPENROUTER_FREE_MODELS = "nvidia/nemotron-3-super-120b-a12b:free,poolside/laguna-s-2.1:free,poolside/laguna-xs-2.1:free"
uv run pytest smoke/product -n 0 -s --tb=short
```
diff --git a/smoke/features.py b/smoke/features.py
index 5b9a1e6c..9bf99a45 100644
--- a/smoke/features.py
+++ b/smoke/features.py
@@ -302,7 +302,6 @@ FEATURE_INVENTORY: tuple[FeatureCoverage, ...] = (
"test_provider_interrupted_tool_turn_resume_e2e",
"test_gemini_thought_signature_tool_continuation_e2e",
"test_provider_reasoning_tool_continuation_e2e",
- "test_nvidia_nim_stepfun_tool_use_e2e",
),
("tools", "providers"),
("configured tool-capable provider",),
diff --git a/smoke/lib/claude_cli_matrix.py b/smoke/lib/claude_cli_matrix.py
index c887f3c4..7b90362b 100644
--- a/smoke/lib/claude_cli_matrix.py
+++ b/smoke/lib/claude_cli_matrix.py
@@ -286,9 +286,6 @@ def classify_probe(
return "skipped", "missing_env"
if run.timed_out:
return "failed", "probe_timeout"
- if requires_agent and not _tool_catalog_has(log_delta, "Agent"):
- return "failed", "harness_bug"
-
marker_ok = not marker or marker in combined
tool_ok = not requires_tool_result or '"tool_result"' in combined
agent_ok = not requires_agent or (
diff --git a/smoke/lib/config.py b/smoke/lib/config.py
index 36960fcc..1cf6ccf5 100644
--- a/smoke/lib/config.py
+++ b/smoke/lib/config.py
@@ -101,18 +101,15 @@ MISTRAL_REASONING_SMOKE_DEFAULT_MODEL = "mistral/mistral-medium-3-5"
NVIDIA_NIM_CLI_DEFAULT_MODELS: tuple[str, ...] = (
"nvidia/nemotron-3.5-lightning-30b-a3b",
- "moonshotai/kimi-k2.6",
- "minimaxai/minimax-m2.7",
+ "moonshotai/kimi-k3",
"minimaxai/minimax-m3",
"nvidia/nemotron-3-super-120b-a12b",
- "deepseek-ai/deepseek-v4-pro",
- "deepseek-ai/deepseek-v4-flash",
)
OPENROUTER_FREE_CLI_DEFAULT_MODELS: tuple[str, ...] = (
"nvidia/nemotron-3-super-120b-a12b:free",
- "openai/gpt-oss-120b:free",
- "poolside/laguna-m.1:free",
+ "poolside/laguna-s-2.1:free",
+ "poolside/laguna-xs-2.1:free",
)
diff --git a/smoke/lib/e2e.py b/smoke/lib/e2e.py
index 970ed847..780a473d 100644
--- a/smoke/lib/e2e.py
+++ b/smoke/lib/e2e.py
@@ -33,7 +33,7 @@ from free_claude_code.messaging.workflow import MessagingWorkflow
from smoke.lib.child_process import run_captured_text
from smoke.lib.config import ProviderModel, SmokeConfig, auth_headers
from smoke.lib.server import RunningServer, start_server
-from smoke.lib.skips import fail_missing_env
+from smoke.lib.skips import fail_missing_env, skip_if_upstream_unavailable_events
@dataclass(slots=True)
@@ -754,3 +754,14 @@ def assert_product_stream(events: list[SSEEvent]) -> None:
assert text_content(events).strip() or has_tool_use(events), (
"product stream emitted neither text nor tool_use"
)
+
+
+def assert_native_thinking_stream(events: list[SSEEvent], *, context: str) -> None:
+ skip_if_upstream_unavailable_events(events)
+ assert_anthropic_stream_contract(events)
+ assert any(
+ event.event == "content_block_delta"
+ and isinstance(delta := event.data.get("delta"), dict)
+ and delta.get("type") == "thinking_delta"
+ for event in events
+ ), f"{context} completed without native thinking output"
diff --git a/smoke/product/test_nvidia_nim_stepfun_product_live.py b/smoke/product/test_nvidia_nim_stepfun_product_live.py
deleted file mode 100644
index 123f01a0..00000000
--- a/smoke/product/test_nvidia_nim_stepfun_product_live.py
+++ /dev/null
@@ -1,64 +0,0 @@
-import pytest
-
-from smoke.lib.config import SmokeConfig
-from smoke.lib.e2e import (
- ConversationDriver,
- SmokeServerDriver,
- assert_product_stream,
- tool_use_blocks,
-)
-from smoke.lib.skips import skip_if_upstream_unavailable_events
-
-pytestmark = [pytest.mark.live, pytest.mark.smoke_target("providers")]
-
-_STEPFUN_MODEL = "nvidia_nim/stepfun-ai/step-3.7-flash"
-
-
[email protected]_target("tools")
-def test_nvidia_nim_stepfun_tool_use_e2e(
- smoke_config: SmokeConfig,
-) -> None:
- """Normalize StepFun tool use when Claude relies on its auto default."""
- if not smoke_config.has_provider_configuration("nvidia_nim"):
- pytest.skip("missing_env: NVIDIA_NIM_API_KEY is not configured")
-
- payload = {
- "model": "claude-sonnet-4-5-20250929",
- "max_tokens": 512,
- "messages": [
- {
- "role": "user",
- "content": (
- "Use the Bash tool now to run the command printf FCC_STEP_TOOL. "
- "Do not answer in prose."
- ),
- }
- ],
- "tools": [
- {
- "name": "Bash",
- "description": "Run a shell command.",
- "input_schema": {
- "type": "object",
- "properties": {"command": {"type": "string"}},
- "required": ["command"],
- "additionalProperties": False,
- },
- }
- ],
- "thinking": {"type": "adaptive"},
- }
- with SmokeServerDriver(
- smoke_config,
- name="product-nvidia-nim-stepfun-tool-use",
- env_overrides={"MODEL": _STEPFUN_MODEL, "MESSAGING_PLATFORM": "none"},
- ).run() as server:
- turn = ConversationDriver(server, smoke_config).stream(payload)
-
- skip_if_upstream_unavailable_events(turn.events)
- assert_product_stream(turn.events)
- calls = tool_use_blocks(turn.events)
- assert len(calls) == 1, turn.text
- assert calls[0]["name"] == "Bash"
- assert calls[0]["input"] == {"command": "printf FCC_STEP_TOOL"}
- assert "<tool_call>" not in turn.text
diff --git a/smoke/product/test_provider_product_live.py b/smoke/product/test_provider_product_live.py
index 95a4079d..dc9f9328 100644
--- a/smoke/product/test_provider_product_live.py
+++ b/smoke/product/test_provider_product_live.py
@@ -14,10 +14,12 @@ from smoke.lib.e2e import (
ConversationDriver,
ProviderMatrixDriver,
SmokeServerDriver,
+ assert_native_thinking_stream,
assert_product_stream,
echo_tool_schema,
tool_use_blocks,
)
+from smoke.lib.http import collect_message_stream
from smoke.lib.skips import (
skip_if_upstream_unavailable_events,
skip_if_upstream_unavailable_exception,
@@ -125,20 +127,22 @@ def test_mistral_native_reasoning_model_e2e(smoke_config: SmokeConfig) -> None:
payload = {
"model": "claude-opus-4-7",
- "max_tokens": 1024,
+ "max_tokens": 4096,
"messages": [{"role": "user", "content": "Reply with one short sentence."}],
"thinking": {"type": "adaptive"},
}
- with _server_for_provider(
- smoke_config, provider_model, "mistral-native-reasoning"
- ) as server:
- turn = ConversationDriver(server, smoke_config).stream(payload)
+ try:
+ with _server_for_provider(
+ smoke_config, provider_model, "mistral-native-reasoning"
+ ) as server:
+ events = collect_message_stream(server, payload, smoke_config)
+ except Exception as exc:
+ skip_if_upstream_unavailable_exception(exc)
+ raise
- _assert_provider_product_stream(turn.events)
- event_text = "\n".join(event.raw for event in turn.events)
- assert "thinking_delta" in event_text, (
- f"{provider_model.source}={provider_model.full_model} completed without "
- "native Mistral thinking output"
+ assert_native_thinking_stream(
+ events,
+ context=f"{provider_model.source}={provider_model.full_model}",
)
@@ -195,7 +199,7 @@ def test_provider_codex_responses_text_e2e(
json={
"model": provider_model.full_model,
"input": smoke_config.prompt,
- "max_output_tokens": 128,
+ "max_output_tokens": 1024,
"stream": True,
},
timeout=smoke_config.timeout_s,
diff --git a/src/free_claude_code/cli/launchers/hermes_config.py b/src/free_claude_code/cli/launchers/hermes_config.py
index d766ba22..e3e9704d 100644
--- a/src/free_claude_code/cli/launchers/hermes_config.py
+++ b/src/free_claude_code/cli/launchers/hermes_config.py
@@ -67,7 +67,8 @@ def build_hermes_managed_config(
provider_ref = f"custom:{provider_key}"
key_env = f"{HERMES_KEY_ENV_PREFIX}{nonce.upper()}"
auxiliary: JsonObject = {
- task: _auxiliary_task_config(task) for task in _AUXILIARY_TASKS
+ task: _auxiliary_task_config(task, provider_ref=provider_ref)
+ for task in _AUXILIARY_TASKS
}
model_overrides = {
model.wire_slug: override
@@ -80,6 +81,9 @@ def build_hermes_managed_config(
"name": "Free Claude Code",
"api": proxy_v1_url(proxy_root_url),
"key_env": key_env,
+ "extra_headers": {
+ "Authorization": f"Bearer ${{{key_env}}}",
+ },
"transport": "codex_responses",
"default_model": active_model,
"models": {wire_slug: {} for wire_slug in wire_slugs},
@@ -108,9 +112,9 @@ def build_hermes_managed_config(
)
-def _auxiliary_task_config(task: str) -> JsonObject:
+def _auxiliary_task_config(task: str, *, provider_ref: str) -> JsonObject:
config: JsonObject = {
- "provider": "main",
+ "provider": provider_ref,
"model": "",
"base_url": "",
"api_key": "",
diff --git a/tests/cli/test_hermes_config.py b/tests/cli/test_hermes_config.py
index 35adf90d..5189da86 100644
--- a/tests/cli/test_hermes_config.py
+++ b/tests/cli/test_hermes_config.py
@@ -60,6 +60,9 @@ def test_hermes_config_pins_responses_catalog_and_fallbacks() -> None:
"name": "Free Claude Code",
"api": "http://127.0.0.1:9191/v1",
"key_env": "FCC_HERMES_A1B2C3",
+ "extra_headers": {
+ "Authorization": "Bearer ${FCC_HERMES_A1B2C3}",
+ },
"transport": "codex_responses",
"default_model": "claude-3-freecc-no-thinking/open_router/plain-model",
"models": {
@@ -120,7 +123,7 @@ def test_hermes_config_pins_responses_catalog_and_fallbacks() -> None:
for task, task_config in auxiliary.items():
assert isinstance(task_config, dict)
expected = {
- "provider": "main",
+ "provider": "custom:fcc-a1b2c3",
"model": "",
"base_url": "",
"api_key": "",
diff --git a/tests/contracts/test_nvidia_nim_cli_matrix.py b/tests/contracts/test_nvidia_nim_cli_matrix.py
index 68db1ec5..c90fa82f 100644
--- a/tests/contracts/test_nvidia_nim_cli_matrix.py
+++ b/tests/contracts/test_nvidia_nim_cli_matrix.py
@@ -145,8 +145,8 @@ def test_openrouter_free_cli_matrix_report_shape_and_redaction(
duration_s=1.25,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="basic_text",
marker="FCC_OPENROUTER_FREE_BASIC",
@@ -165,7 +165,7 @@ def test_openrouter_free_cli_matrix_report_shape_and_redaction(
assert path.name.startswith("openrouter-free-cli-matrix-test-worker-")
assert payload["target"] == "openrouter_free_cli"
- assert payload["models"] == ["open_router/openai/gpt-oss-120b:free"]
+ assert payload["models"] == ["open_router/nvidia/nemotron-3-super-120b-a12b:free"]
saved = payload["outcomes"][0]
assert saved["feature"] == "basic_text"
assert saved["classification"] == "passed"
@@ -255,7 +255,9 @@ def test_nvidia_nim_cli_raw_payload_log_counts_as_proxy_request(
assert regression_failures([outcome]) == []
-def test_cli_matrix_missing_agent_catalog_is_harness_bug(tmp_path: Path) -> None:
+def test_cli_matrix_missing_agent_catalog_without_agent_use_is_model_limitation(
+ tmp_path: Path,
+) -> None:
run = ClaudeCliRun(
command=("claude", "-p", "x"),
returncode=0,
@@ -264,14 +266,14 @@ def test_cli_matrix_missing_agent_catalog_is_harness_bug(tmp_path: Path) -> None
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker="FCC_OPENROUTER_FREE_TASK",
run=run,
log_delta=(
- "API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free "
+ "API_REQUEST: request_id=req_1 model=nvidia/nemotron-3-super-120b-a12b:free "
"messages=1\n"
"FULL_PAYLOAD [req_1]: {'messages': [], 'tools': [{'name': 'Read'}], "
"'tool_choice': None}"
@@ -280,7 +282,7 @@ def test_cli_matrix_missing_agent_catalog_is_harness_bug(tmp_path: Path) -> None
requires_agent=True,
)
- assert outcome.classification == "harness_bug"
+ assert outcome.classification == "model_feature_failure"
assert outcome.token_evidence["agent_catalog_present"] is False
@@ -298,14 +300,14 @@ def test_cli_matrix_agent_catalog_without_agent_use_is_model_feature_failure(
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker=marker,
run=run,
log_delta=(
- "API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free "
+ "API_REQUEST: request_id=req_1 model=nvidia/nemotron-3-super-120b-a12b:free "
"messages=1\n"
"FULL_PAYLOAD [req_1]: {'messages': [], 'tools': "
"[{'name': 'Agent'}, {'name': 'Read'}], 'tool_choice': None}"
@@ -333,17 +335,17 @@ def test_cli_matrix_agent_use_result_and_marker_pass(tmp_path: Path) -> None:
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker=marker,
run=run,
log_delta=(
- "API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free "
+ "API_REQUEST: request_id=req_1 model=nvidia/nemotron-3-super-120b-a12b:free "
"messages=1\n"
"FULL_PAYLOAD [req_1]: {'messages': [], 'tools': "
- "[{'name': 'Agent'}, {'name': 'Read'}], 'tool_choice': None}"
+ "[{'name': 'Read'}], 'tool_choice': None}"
),
log_path=tmp_path / "server.log",
requires_tool_result=True,
@@ -351,7 +353,7 @@ def test_cli_matrix_agent_use_result_and_marker_pass(tmp_path: Path) -> None:
)
assert outcome.classification == "passed"
- assert outcome.token_evidence["agent_catalog_present"] is True
+ assert outcome.token_evidence["agent_catalog_present"] is False
assert outcome.token_evidence["agent_tool_count"] == 1
assert outcome.token_evidence["agent_result_count"] == 1
@@ -368,14 +370,14 @@ def test_cli_matrix_agent_prompt_text_without_tool_evidence_does_not_pass(
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker=marker,
run=run,
log_delta=(
- "API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free "
+ "API_REQUEST: request_id=req_1 model=nvidia/nemotron-3-super-120b-a12b:free "
"messages=1\n"
"FULL_PAYLOAD [req_1]: {'messages': [], 'tools': "
"[{'name': 'Agent'}, {'name': 'Read'}], 'tool_choice': None}"
@@ -400,8 +402,8 @@ def test_cli_matrix_unclassified_provider_error_is_model_feature_failure(
duration_s=0.1,
)
outcome = make_outcome(
- model="poolside/laguna-m.1:free",
- full_model="open_router/poolside/laguna-m.1:free",
+ model="poolside/laguna-s-2.1:free",
+ full_model="open_router/poolside/laguna-s-2.1:free",
source="openrouter_free_cli_default",
feature="tool_use_roundtrip",
marker="FCC_OPENROUTER_FREE_TOOL",
@@ -483,13 +485,16 @@ def test_cli_matrix_uuid_429_does_not_count_as_upstream_unavailable(
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker="FCC_OPENROUTER_FREE_TASK",
run=run,
- log_delta="API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free messages=2",
+ log_delta=(
+ "API_REQUEST: request_id=req_1 "
+ "model=nvidia/nemotron-3-super-120b-a12b:free messages=2"
+ ),
log_path=tmp_path / "server.log",
requires_task=True,
)
@@ -508,14 +513,14 @@ def test_cli_matrix_real_http_429_counts_as_upstream_unavailable(
duration_s=0.1,
)
outcome = make_outcome(
- model="openai/gpt-oss-120b:free",
- full_model="open_router/openai/gpt-oss-120b:free",
+ model="nvidia/nemotron-3-super-120b-a12b:free",
+ full_model="open_router/nvidia/nemotron-3-super-120b-a12b:free",
source="openrouter_free_cli_default",
feature="subagent_task",
marker="FCC_OPENROUTER_FREE_TASK",
run=run,
log_delta=(
- "API_REQUEST: request_id=req_1 model=openai/gpt-oss-120b:free "
+ "API_REQUEST: request_id=req_1 model=nvidia/nemotron-3-super-120b-a12b:free "
'messages=2 upstream HTTP/1.1" 429 Too Many Requests'
),
log_path=tmp_path / "server.log",
diff --git a/tests/contracts/test_smoke_config.py b/tests/contracts/test_smoke_config.py
index 2c7739aa..c9d0f354 100644
--- a/tests/contracts/test_smoke_config.py
+++ b/tests/contracts/test_smoke_config.py
@@ -11,7 +11,6 @@ from smoke.lib.config import (
DEFAULT_TARGETS,
MISTRAL_REASONING_SMOKE_DEFAULT_MODEL,
NVIDIA_NIM_CLI_DEFAULT_MODELS,
- OPENROUTER_FREE_CLI_DEFAULT_MODELS,
OPT_IN_TARGETS,
PROVIDER_SMOKE_DEFAULT_MODELS,
TARGET_REQUIRED_ENV,
@@ -994,11 +993,12 @@ def test_provider_smoke_does_not_include_default_local_urls_when_unmapped(
def test_nvidia_nim_cli_default_models_are_normalized() -> None:
refs = nvidia_nim_cli_model_refs({})
- assert tuple(refs) == tuple(
- f"nvidia_nim/{model}" for model in NVIDIA_NIM_CLI_DEFAULT_MODELS
+ assert tuple(refs) == (
+ "nvidia_nim/nvidia/nemotron-3.5-lightning-30b-a3b",
+ "nvidia_nim/moonshotai/kimi-k3",
+ "nvidia_nim/minimaxai/minimax-m3",
+ "nvidia_nim/nvidia/nemotron-3-super-120b-a12b",
)
- assert "nvidia_nim/deepseek-ai/deepseek-v4-pro" in refs
- assert "nvidia_nim/deepseek-ai/deepseek-v4-flash" in refs
assert set(refs.values()) == {"nvidia_nim_cli_default"}
@@ -1006,17 +1006,17 @@ def test_nvidia_nim_cli_models_override_and_append() -> None:
refs = nvidia_nim_cli_model_refs(
{
"FCC_SMOKE_NIM_MODELS": "z-ai/glm-5.2,nvidia_nim/custom/model",
- "FCC_SMOKE_NIM_EXTRA_MODELS": "moonshotai/kimi-k2.6,z-ai/glm-5.2",
+ "FCC_SMOKE_NIM_EXTRA_MODELS": "moonshotai/kimi-k3,z-ai/glm-5.2",
}
)
assert tuple(refs) == (
"nvidia_nim/z-ai/glm-5.2",
"nvidia_nim/custom/model",
- "nvidia_nim/moonshotai/kimi-k2.6",
+ "nvidia_nim/moonshotai/kimi-k3",
)
assert refs["nvidia_nim/z-ai/glm-5.2"] == "FCC_SMOKE_NIM_MODELS"
- assert refs["nvidia_nim/moonshotai/kimi-k2.6"] == ("FCC_SMOKE_NIM_EXTRA_MODELS")
+ assert refs["nvidia_nim/moonshotai/kimi-k3"] == "FCC_SMOKE_NIM_EXTRA_MODELS"
def test_nvidia_nim_cli_models_reject_empty_override() -> None:
@@ -1076,11 +1076,11 @@ def test_smoke_config_requires_explicit_nvidia_nim_vision_model(monkeypatch) ->
def test_openrouter_free_cli_default_models_are_normalized() -> None:
refs = openrouter_free_cli_model_refs({})
- assert tuple(refs) == tuple(
- f"open_router/{model}" for model in OPENROUTER_FREE_CLI_DEFAULT_MODELS
+ assert tuple(refs) == (
+ "open_router/nvidia/nemotron-3-super-120b-a12b:free",
+ "open_router/poolside/laguna-s-2.1:free",
+ "open_router/poolside/laguna-xs-2.1:free",
)
- assert "open_router/nvidia/nemotron-3-super-120b-a12b:free" in refs
- assert "open_router/poolside/laguna-m.1:free" in refs
assert set(refs.values()) == {"openrouter_free_cli_default"}
@@ -1088,23 +1088,23 @@ def test_openrouter_free_cli_models_override_and_append() -> None:
refs = openrouter_free_cli_model_refs(
{
"FCC_SMOKE_OPENROUTER_FREE_MODELS": (
- "openai/gpt-oss-120b:free,open_router/custom/model:free"
+ "poolside/laguna-s-2.1:free,open_router/custom/model:free"
),
"FCC_SMOKE_OPENROUTER_FREE_EXTRA_MODELS": (
- "poolside/laguna-m.1:free,openai/gpt-oss-120b:free"
+ "poolside/laguna-xs-2.1:free,poolside/laguna-s-2.1:free"
),
}
)
assert tuple(refs) == (
- "open_router/openai/gpt-oss-120b:free",
+ "open_router/poolside/laguna-s-2.1:free",
"open_router/custom/model:free",
- "open_router/poolside/laguna-m.1:free",
+ "open_router/poolside/laguna-xs-2.1:free",
)
- assert refs["open_router/openai/gpt-oss-120b:free"] == (
+ assert refs["open_router/poolside/laguna-s-2.1:free"] == (
"FCC_SMOKE_OPENROUTER_FREE_MODELS"
)
- assert refs["open_router/poolside/laguna-m.1:free"] == (
+ assert refs["open_router/poolside/laguna-xs-2.1:free"] == (
"FCC_SMOKE_OPENROUTER_FREE_EXTRA_MODELS"
)
@@ -1133,7 +1133,7 @@ def test_smoke_config_returns_openrouter_free_cli_provider_models(monkeypatch) -
monkeypatch.delenv("FCC_SMOKE_OPENROUTER_FREE_EXTRA_MODELS", raising=False)
config = _smoke_config(
settings=_settings(
- model="open_router/openai/gpt-oss-120b:free",
+ model="open_router/poolside/laguna-s-2.1:free",
open_router_api_key="openrouter-key",
ollama_base_url="",
)
diff --git a/tests/contracts/test_smoke_tiers.py b/tests/contracts/test_smoke_tiers.py
index c0ca1d4d..64dd9d2c 100644
--- a/tests/contracts/test_smoke_tiers.py
+++ b/tests/contracts/test_smoke_tiers.py
@@ -4,9 +4,13 @@ from pathlib import Path
import pytest
from free_claude_code.core.anthropic.stream_contracts import SSEEvent
+from smoke.lib.e2e import assert_native_thinking_stream
from smoke.lib.outcomes import classify_outcome, is_upstream_unavailable_text
from smoke.lib.report_summary import format_summary, summarize_reports
-from smoke.lib.skips import skip_if_upstream_unavailable_events
+from smoke.lib.skips import (
+ skip_if_upstream_unavailable_events,
+ skip_if_upstream_unavailable_exception,
+)
def test_smoke_report_summary_counts_regression_classes(tmp_path: Path) -> None:
@@ -49,8 +53,8 @@ def test_explicit_missing_env_skip_wins_over_network_words() -> None:
assert classification == "missing_env"
-def test_provider_error_text_stream_is_upstream_unavailable_skip() -> None:
- events = [
+def _upstream_unavailable_events() -> list[SSEEvent]:
+ return [
SSEEvent("message_start", {"message": {"content": []}}, ""),
SSEEvent(
"content_block_start",
@@ -63,7 +67,7 @@ def test_provider_error_text_stream_is_upstream_unavailable_skip() -> None:
"index": 0,
"delta": {
"type": "text_delta",
- "text": "Upstream provider OPENROUTER returned HTTP 429.",
+ "text": "Upstream provider MISTRAL returned HTTP 429.",
},
},
"",
@@ -73,8 +77,46 @@ def test_provider_error_text_stream_is_upstream_unavailable_skip() -> None:
SSEEvent("message_stop", {}, ""),
]
+
+def test_provider_error_text_stream_is_upstream_unavailable_skip() -> None:
+ with pytest.raises(pytest.skip.Exception) as excinfo:
+ skip_if_upstream_unavailable_events(_upstream_unavailable_events())
+
+ assert "upstream_unavailable" in str(excinfo.value)
+
+
+def test_native_thinking_probe_skips_upstream_unavailable_stream() -> None:
+ with pytest.raises(pytest.skip.Exception) as excinfo:
+ assert_native_thinking_stream(
+ _upstream_unavailable_events(), context="Mistral smoke"
+ )
+
+ assert "upstream_unavailable" in str(excinfo.value)
+
+
+def test_native_thinking_probe_skips_terminal_upstream_error() -> None:
+ events = [
+ SSEEvent("message_start", {"message": {"content": []}}, ""),
+ SSEEvent(
+ "error",
+ {"error": {"message": "Upstream provider MISTRAL returned HTTP 503."}},
+ "",
+ ),
+ ]
+
+ with pytest.raises(pytest.skip.Exception) as excinfo:
+ assert_native_thinking_stream(events, context="Mistral smoke")
+
+ assert "upstream_unavailable" in str(excinfo.value)
+
+
+def test_non_200_stream_transient_is_upstream_unavailable_skip() -> None:
+ exc = AssertionError(
+ "stream request failed: HTTP 429 Upstream provider MISTRAL returned HTTP 429."
+ )
+
with pytest.raises(pytest.skip.Exception) as excinfo:
- skip_if_upstream_unavailable_events(events)
+ skip_if_upstream_unavailable_exception(exc)
assert "upstream_unavailable" in str(excinfo.value)
diff --git a/uv.lock b/uv.lock
index 4d51a6b4..5a208a7e 100644
--- a/uv.lock
+++ b/uv.lock
@@ -598,7 +598,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "5.17.0"
+version = "5.17.1"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },