Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Tests: price the auto-context fit floor as the layer-split path on any host (#16014)
Daniel Han8ecab3dbbb20
diff --git a/studio/backend/tests/test_memory_estimate.py b/studio/backend/tests/test_memory_estimate.pyindex 7d6ff7908f9..7dbec10a12e 100644--- a/studio/backend/tests/test_memory_estimate.py+++ b/studio/backend/tests/test_memory_estimate.py@@ -2871,6 +2871,11 @@ class TestAnAutoContextIsMarkedShrinkable:@pytest.fixture(autouse = True)def wide(self, tmp_path, monkeypatch):+ import utils.hardware as hardware++ # The layer-split floor (8192). Metal reprices down to 256 instead; that path is+ # TestTheContextFloorFollowsThePlacementPath's, and on a Mac runner it would be taken here.+ monkeypatch.setattr(hardware, "is_apple_silicon", lambda: False)gguf = _write_gguf(tmp_path, "qwen3", {**_GQA_FIELDS, "context_length": 262144})config = SimpleNamespace(identifier = "local/wide",