diff --git a/tests/e2e/test_model_alias_routing.py b/tests/e2e/test_model_alias_routing.py index f20c24af..8ad71179 100644 --- a/tests/e2e/test_model_alias_routing.py +++ b/tests/e2e/test_model_alias_routing.py @@ -480,6 +480,7 @@ def test_anthropic_thinking_mode_streaming(): assert "thinking" in final_block_types +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_client_with_coding_model_alias_and_tools(): """Test OpenAI client using 'coding-model' alias (maps to Bedrock) with coding question and tools""" logger.info("Testing OpenAI client with 'coding-model' alias -> Bedrock with tools") @@ -529,6 +530,7 @@ def test_openai_client_with_coding_model_alias_and_tools(): ) +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_anthropic_client_with_coding_model_alias_and_tools(): """Test Anthropic client using 'coding-model' alias (maps to Bedrock) with coding question and tools""" logger.info( @@ -576,7 +578,7 @@ def test_anthropic_client_with_coding_model_alias_and_tools(): assert text_content or len(tool_use_blocks) > 0 -@pytest.mark.skip("flay test - to be fixed") +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_anthropic_client_with_coding_model_alias_and_tools_streaming(): """Test Anthropic client using 'coding-model' alias (maps to Bedrock) with coding question and tools - streaming""" logger.info( @@ -671,6 +673,7 @@ def test_anthropic_client_with_coding_model_alias_and_tools_streaming(): ), f"Final message should have content. Got: {final_message.content if final_message else 'None'}" +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_anthropic_client_streaming_with_bedrock(): """Test Anthropic client using 'coding-model' alias (maps to Bedrock) with streaming""" logger.info( @@ -711,6 +714,7 @@ def test_anthropic_client_streaming_with_bedrock(): assert final_message.content and len(final_message.content) > 0 +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_client_streaming_with_bedrock(): """Test OpenAI client using 'coding-model' alias (maps to Bedrock) with streaming""" logger.info( @@ -749,6 +753,7 @@ def test_openai_client_streaming_with_bedrock(): assert len(full_content) > 0, "Expected text response from streaming" +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_client_streaming_with_bedrock_and_tools(): """Test OpenAI client using 'coding-model' alias (maps to Bedrock) with streaming and tools""" logger.info( diff --git a/tests/e2e/test_openai_responses_api_client.py b/tests/e2e/test_openai_responses_api_client.py index 33f91be3..6e110e0d 100644 --- a/tests/e2e/test_openai_responses_api_client.py +++ b/tests/e2e/test_openai_responses_api_client.py @@ -327,6 +327,7 @@ def test_openai_responses_api_streaming_with_tools_upstream_chat_completions(): ), "Expected streamed text or tool call argument deltas from Responses tools stream" +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_responses_api_non_streaming_upstream_bedrock(): """Send a v1/responses request using the coding-model alias to verify Bedrock translation/routing""" base_url = LLM_GATEWAY_ENDPOINT.replace("/v1/chat/completions", "") @@ -347,6 +348,7 @@ def test_openai_responses_api_non_streaming_upstream_bedrock(): assert resp.id is not None +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_responses_api_with_streaming_upstream_bedrock(): """Build a v1/responses API streaming request routed to Bedrock via coding-model alias""" base_url = LLM_GATEWAY_ENDPOINT.replace("/v1/chat/completions", "") @@ -392,6 +394,7 @@ def test_openai_responses_api_with_streaming_upstream_bedrock(): assert len(full_content) > 0, "Should have received content" +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_responses_api_non_streaming_with_tools_upstream_bedrock(): """Responses API with tools routed to Bedrock via coding-model alias""" base_url = LLM_GATEWAY_ENDPOINT.replace("/v1/chat/completions", "") @@ -424,6 +427,7 @@ def test_openai_responses_api_non_streaming_with_tools_upstream_bedrock(): print(f"{'='*80}\n") +@pytest.mark.skip("unreliable - bedrock tests are flaky in CI") def test_openai_responses_api_streaming_with_tools_upstream_bedrock(): """Responses API with a function/tool definition streaming to Bedrock via coding-model alias""" base_url = LLM_GATEWAY_ENDPOINT.replace("/v1/chat/completions", "")