CHANGELOG.md
CONTRIBUTING.md
LICENSE
LICENSE-MIT
MANIFEST.in
README.md
pyproject.toml
bench/README.md
bench/example.json
bench/merge.py
bench/plot-bench.py
bench/serve-bench.py
bench/patches/ds4-ignore-eos.patch
docs/README.md
docs/adding-architectures.md
docs/arch-coverage.md
docs/assistant.md
docs/benchmarks.json
docs/benchmarks.md
docs/cli.md
docs/getting-started.md
docs/launch.md
docs/lora.md
docs/migrating.md
docs/performance.md
docs/python.md
docs/rag.md
docs/server-config.md
docs/serving-architecture.md
docs/speculative-batching.md
docs/streaming.md
docs/talk.md
docs/testing.md
docs/troubleshooting.md
docs/upstream-upgrades.md
docs/vlm.md
docs/assets/demo.gif
docs/assets/record-demo.sh
docs/assets/perf/fleet-ratio-dark.svg
docs/assets/perf/fleet-ratio.svg
docs/assets/perf/kimi-k3-ud-iq2xxs-car.html
docs/assets/perf/kimi-k3-ud-iq2xxs-car.png
docs/assets/perf/kimi-k3-ud-q2kxl-car.html
docs/assets/perf/kimi-k3-ud-q2kxl-car.png
docs/assets/perf/lossy-hy3-baseline.html
docs/assets/perf/lossy-hy3-baseline.png
docs/assets/perf/lossy-hy3-shed-0.07-0.93.html
docs/assets/perf/lossy-hy3-shed-0.07-0.93.png
docs/assets/perf/lossy-hy3-shed-0.10-0.90-cool.html
docs/assets/perf/lossy-hy3-shed-0.10-0.90-cool.png
docs/assets/perf/lossy-hy3-shed-0.10-0.90.html
docs/assets/perf/lossy-hy3-shed-0.10-0.90.png
docs/assets/perf/lossy-hy3-shed-0.20-0.80.html
docs/assets/perf/lossy-hy3-shed-0.20-0.80.png
docs/assets/perf/lossy-kimi-k3-shed-0.65.html
docs/assets/perf/lossy-kimi-k3-shed-0.65.png
docs/assets/perf/lossy-kimi-k3-shed-0.70.html
docs/assets/perf/lossy-kimi-k3-shed-0.70.png
docs/assets/perf/lossy-kimi-k3-shed-0.80.html
docs/assets/perf/lossy-kimi-k3-shed-0.80.png
docs/assets/perf/mtp-lift-dark.svg
docs/assets/perf/mtp-lift.svg
docs/assets/perf/per-model/deepseek-v4-flash-antirez-iq2xxs-panels-dark.svg
docs/assets/perf/per-model/deepseek-v4-flash-antirez-iq2xxs-panels.svg
docs/assets/perf/per-model/deepseek-v4-flash-unsloth-udiq3xxs-panels-dark.svg
docs/assets/perf/per-model/deepseek-v4-flash-unsloth-udiq3xxs-panels.svg
docs/assets/perf/per-model/dolphin3-llama3.1-8b-q6k-panels-dark.svg
docs/assets/perf/per-model/dolphin3-llama3.1-8b-q6k-panels.svg
docs/assets/perf/per-model/gemma-4-12b-q6k-panels-dark.svg
docs/assets/perf/per-model/gemma-4-12b-q6k-panels.svg
docs/assets/perf/per-model/gemma-4-26b-a4b-q6k-panels-dark.svg
docs/assets/perf/per-model/gemma-4-26b-a4b-q6k-panels.svg
docs/assets/perf/per-model/gemma-4-31b-q6k-panels-dark.svg
docs/assets/perf/per-model/gemma-4-31b-q6k-panels.svg
docs/assets/perf/per-model/gemma-4-e2b-q6kxl-panels-dark.svg
docs/assets/perf/per-model/gemma-4-e2b-q6kxl-panels.svg
docs/assets/perf/per-model/gemma-4-e4b-q6k-panels-dark.svg
docs/assets/perf/per-model/gemma-4-e4b-q6k-panels.svg
docs/assets/perf/per-model/gpt-oss-120b-heretic-mxfp4-panels-dark.svg
docs/assets/perf/per-model/gpt-oss-120b-heretic-mxfp4-panels.svg
docs/assets/perf/per-model/gpt-oss-20b-mxfp4-panels-dark.svg
docs/assets/perf/per-model/gpt-oss-20b-mxfp4-panels.svg
docs/assets/perf/per-model/qwen3.5-122b-a10b-ud-q5km-panels-dark.svg
docs/assets/perf/per-model/qwen3.5-122b-a10b-ud-q5km-panels.svg
docs/assets/perf/per-model/qwen3.5-9b-q6k-panels-dark.svg
docs/assets/perf/per-model/qwen3.5-9b-q6k-panels.svg
docs/assets/perf/per-model/qwen3.6-27b-q6k-panels-dark.svg
docs/assets/perf/per-model/qwen3.6-27b-q6k-panels.svg
docs/assets/perf/per-model/qwen3.6-35b-a3b-heretic-q6k-panels-dark.svg
docs/assets/perf/per-model/qwen3.6-35b-a3b-heretic-q6k-panels.svg
gmlx/__init__.py
gmlx/__main__.py
gmlx/adapter.py
gmlx/admit_gate.py
gmlx/apc_manager.py
gmlx/apc_pooling.py
gmlx/arch_table.py
gmlx/arrays_cache_fix.py
gmlx/assistant_brain.py
gmlx/assistant_serve.py
gmlx/attn_hd512.py
gmlx/auto_ratio.py
gmlx/batch_sched.py
gmlx/benchmarks.py
gmlx/cache_compat.py
gmlx/cache_snapshot.py
gmlx/cascade_sdpa.py
gmlx/cb_phase.py
gmlx/chat.py
gmlx/cli.py
gmlx/completion.py
gmlx/config.py
gmlx/config_synth.py
gmlx/decode_feeder.py
gmlx/deepseek_v4_cache.py
gmlx/deepseek_v4_dspark.py
gmlx/deepseek_v4_hyper_connection.py
gmlx/deepseek_v4_model.py
gmlx/deepseek_v4_mtp.py
gmlx/diffusion.py
gmlx/discovery.py
gmlx/doctor.py
gmlx/drafter_protocol.py
gmlx/dsv32_patches.py
gmlx/dtypes.py
gmlx/embeddings.py
gmlx/envflags.py
gmlx/extras.py
gmlx/feeder.py
gmlx/feeder_common.py
gmlx/gdn_patches.py
gmlx/gemma4_batched_sdpa.py
gmlx/gemma4_owned.py
gmlx/gemma4_sync.py
gmlx/generation.py
gmlx/gguf_meta.py
gmlx/gpt_oss_prefill.py
gmlx/gpu_token.py
gmlx/headerscan.py
gmlx/hf_cache.py
gmlx/hotkey.py
gmlx/hy_v3_model.py
gmlx/hy_v3_mtp.py
gmlx/hy_v3_tools.py
gmlx/keepwarm.py
gmlx/kimi_k3_model.py
gmlx/launch.py
gmlx/lifecycle.py
gmlx/loader.py
gmlx/loadlog.py
gmlx/lookahead.py
gmlx/manage.py
gmlx/memfit.py
gmlx/menubar.py
gmlx/menubar_config.py
gmlx/minimax_m3_model.py
gmlx/modules.py
gmlx/moe_experts.py
gmlx/mtp_drafter.py
gmlx/mtp_load.py
gmlx/muse_glimmer_dflash.py
gmlx/muse_glimmer_model.py
gmlx/muse_glimmer_mtp.py
gmlx/muse_glimmer_tools.py
gmlx/muse_glimmer_vlm_model.py
gmlx/native_fp.py
gmlx/occupancy_fuse.py
gmlx/overgen.py
gmlx/pagecache.py
gmlx/patching.py
gmlx/pin_weights.py
gmlx/populate.py
gmlx/prefetch.py
gmlx/prefill_decay.py
gmlx/prefix_cache.py
gmlx/preflight.py
gmlx/procname.py
gmlx/profiles.py
gmlx/qkv_fuse.py
gmlx/quantized_sdpa_fix.py
gmlx/qwen35_attn.py
gmlx/qwen35_gdn.py
gmlx/qwen35_layers.py
gmlx/qwen35_owned.py
gmlx/qwen35_rope.py
gmlx/qwen35_verify_fold.py
gmlx/qwen35_verify_linear.py
gmlx/ragged_decode.py
gmlx/reasoning.py
gmlx/remap.py
gmlx/remote.py
gmlx/render.py
gmlx/rerank.py
gmlx/residency.py
gmlx/retire_key.py
gmlx/rope_batch_fix.py
gmlx/rotating_cache_fix.py
gmlx/serve_memtrace.py
gmlx/server.py
gmlx/server_bridge_lm.py
gmlx/server_bridge_vlm.py
gmlx/server_memory.py
gmlx/sessions.py
gmlx/sparse_sdpa.py
gmlx/spec_engine.py
gmlx/spec_helpers.py
gmlx/speculative.py
gmlx/spinner.py
gmlx/step_timing.py
gmlx/stt.py
gmlx/subservice.py
gmlx/talk.py
gmlx/talk_audio.py
gmlx/talk_client.py
gmlx/talk_mcp.py
gmlx/talk_memory.py
gmlx/textfmt.py
gmlx/theme.py
gmlx/thinking_budget.py
gmlx/tokenizer.py
gmlx/train.py
gmlx/transforms.py
gmlx/tts.py
gmlx/upstream_seams.json
gmlx/upstream_seams.py
gmlx/vlm.py
gmlx/wizard.py
gmlx.egg-info/PKG-INFO
gmlx.egg-info/SOURCES.txt
gmlx.egg-info/dependency_links.txt
gmlx.egg-info/entry_points.txt
gmlx.egg-info/requires.txt
gmlx.egg-info/top_level.txt
gmlx/_vendor/__init__.py
gmlx/_vendor/misaki/LICENSE
gmlx/_vendor/misaki/__init__.py
gmlx/_vendor/misaki/cutlet.py
gmlx/_vendor/misaki/en.py
gmlx/_vendor/misaki/espeak.py
gmlx/_vendor/misaki/he.py
gmlx/_vendor/misaki/ja.py
gmlx/_vendor/misaki/ko.py
gmlx/_vendor/misaki/num2kana.py
gmlx/_vendor/misaki/token.py
gmlx/_vendor/misaki/tone_sandhi.py
gmlx/_vendor/misaki/transcription.py
gmlx/_vendor/misaki/vi.py
gmlx/_vendor/misaki/zh.py
gmlx/_vendor/misaki/zh_frontend.py
gmlx/_vendor/misaki/data/__init__.py
gmlx/_vendor/misaki/data/gb_gold.json
gmlx/_vendor/misaki/data/gb_silver.json
gmlx/_vendor/misaki/data/ja_words.txt
gmlx/_vendor/misaki/data/us_gold.json
gmlx/_vendor/misaki/data/us_silver.json
gmlx/_vendor/misaki/data/vi_acronyms.json
gmlx/_vendor/misaki/data/vi_symbols.json
gmlx/_vendor/misaki/data/vi_teencode.json
gmlx/_vendor/misaki/g2pkc/__init__.py
gmlx/_vendor/misaki/g2pkc/english.py
gmlx/_vendor/misaki/g2pkc/g2pk.py
gmlx/_vendor/misaki/g2pkc/idioms.txt
gmlx/_vendor/misaki/g2pkc/numerals.py
gmlx/_vendor/misaki/g2pkc/regular.py
gmlx/_vendor/misaki/g2pkc/rules.txt
gmlx/_vendor/misaki/g2pkc/special.py
gmlx/_vendor/misaki/g2pkc/table.csv
gmlx/_vendor/misaki/g2pkc/utils.py
gmlx/_vendor/misaki/vi_cleaner/__init__.py
gmlx/_vendor/misaki/vi_cleaner/abbreviation_vi.py
gmlx/_vendor/misaki/vi_cleaner/acronym_vi.py
gmlx/_vendor/misaki/vi_cleaner/cleaner.py
gmlx/_vendor/misaki/vi_cleaner/currency_vi.py
gmlx/_vendor/misaki/vi_cleaner/datestime_vi.py
gmlx/_vendor/misaki/vi_cleaner/letter_vi.py
gmlx/_vendor/misaki/vi_cleaner/measurement_vi.py
gmlx/_vendor/misaki/vi_cleaner/num2vi.py
gmlx/_vendor/misaki/vi_cleaner/numberical_vi.py
gmlx/_vendor/misaki/vi_cleaner/passage_utils.py
gmlx/_vendor/misaki/vi_cleaner/roman_number_vi.py
gmlx/_vendor/misaki/vi_cleaner/sentence_utils.py
gmlx/_vendor/misaki/vi_cleaner/symbol_vi.py
gmlx/_vendor/misaki/zh_normalization/__init__.py
gmlx/_vendor/misaki/zh_normalization/char_convert.py
gmlx/_vendor/misaki/zh_normalization/chronology.py
gmlx/_vendor/misaki/zh_normalization/constants.py
gmlx/_vendor/misaki/zh_normalization/num.py
gmlx/_vendor/misaki/zh_normalization/phonecode.py
gmlx/_vendor/misaki/zh_normalization/quantifier.py
gmlx/_vendor/misaki/zh_normalization/text_normalization.py
gmlx/server_patches/__init__.py
gmlx/server_patches/_common.py
gmlx/server_patches/apc.py
gmlx/server_patches/api_contract.py
gmlx/server_patches/chat_behavior.py
gmlx/server_patches/completions.py
gmlx/server_patches/hardening.py
gmlx/server_patches/observability.py
gmlx/server_patches/render.py
gmlx/server_patches/request_flow.py
gmlx/server_patches/routes.py
gmlx/server_patches/sampling.py
licenses/deepseek-v4-LICENSE
licenses/ds4.c-LICENSE
licenses/llama.cpp-LICENSE
licenses/misaki-LICENSE
licenses/mlx-lm-LICENSE
licenses/mlx-vlm-LICENSE
licenses/omlx-LICENSE
tests/conftest.py
tests/test_activation_dtype.py
tests/test_adapter.py
tests/test_adapter_save.py
tests/test_admit_gate.py
tests/test_apc_ckpt_stats.py
tests/test_apc_manager.py
tests/test_apc_pooling.py
tests/test_apc_routing.py
tests/test_apc_schedule_invariant.py
tests/test_apc_store_eval.py
tests/test_api_contract.py
tests/test_arch_gate.py
tests/test_arrays_cache_fix.py
tests/test_ascii_hygiene.py
tests/test_assistant_brain.py
tests/test_assistant_serve.py
tests/test_attn_hd512_verify.py
tests/test_auto_ratio.py
tests/test_batch_parity.py
tests/test_batch_sched.py
tests/test_bench_prompt_source.py
tests/test_cache_snapshot.py
tests/test_cascade_sdpa.py
tests/test_cb_phase.py
tests/test_chat.py
tests/test_chat_assistant.py
tests/test_chat_e2e.py
tests/test_chat_pty.py
tests/test_chat_sessions.py
tests/test_ckpt_cursor.py
tests/test_ckpt_decode_lcp.py
tests/test_ckpt_plain_path.py
tests/test_ckpt_pool_pressure.py
tests/test_ckpt_tier.py
tests/test_cli_config_model.py
tests/test_cli_consistency.py
tests/test_cli_run_flags.py
tests/test_cli_umbrella.py
tests/test_cli_validation.py
tests/test_cmd_profiles.py
tests/test_completion.py
tests/test_completions_route.py
tests/test_config.py
tests/test_config_synth.py
tests/test_decode_feeder.py
tests/test_deepseek_v4_hyper_connection.py
tests/test_deepseek_v4_indexer_quant.py
tests/test_deepseek_v4_kv_quant.py
tests/test_deepseek_v4_mtp.py
tests/test_deepseek_v4_qat.py
tests/test_deepseek_v4_score_profile.py
tests/test_dequant_unattachable.py
tests/test_dflash_normalize.py
tests/test_diffusion_gemma.py
tests/test_discovery.py
tests/test_docs_config.py
tests/test_docs_python.py
tests/test_doctor.py
tests/test_draft_block_depth.py
tests/test_drafter_protocol.py
tests/test_ds4_overlap_lookback.py
tests/test_dsv32_indexer_rope.py
tests/test_dsv32_mask_decode.py
tests/test_e2e_checks.py
tests/test_e2e_harness_smoke.py
tests/test_embeddings.py
tests/test_envflags.py
tests/test_exact_anchor.py
tests/test_extras_pyproject.py
tests/test_faithful_history.py
tests/test_feeder.py
tests/test_forward_contract_pins.py
tests/test_full_prompt_prefill.py
tests/test_gdn_ba_cat.py
tests/test_gdn_noop_mask.py
tests/test_gemma4_batched_sdpa.py
tests/test_gemma4_owned.py
tests/test_gemma4_sync.py
tests/test_generate_over.py
tests/test_gpu_token.py
tests/test_headerscan.py
tests/test_hf_cache.py
tests/test_hotkey.py
tests/test_hy_v3_mtp.py
tests/test_hy_v3_tools.py
tests/test_hyv3_shexp_fold.py
tests/test_init_wizard.py
tests/test_keepwarm.py
tests/test_kimi_k2.py
tests/test_kimi_k25_vision.py
tests/test_kimi_k3.py
tests/test_kimi_k3_template.py
tests/test_launch.py
tests/test_lifecycle.py
tests/test_loader_api.py
tests/test_loader_guards.py
tests/test_loadlog.py
tests/test_long_context.py
tests/test_lookahead.py
tests/test_manage.py
tests/test_memfit.py
tests/test_menubar.py
tests/test_menubar_config.py
tests/test_minimax_m3_msa.py
tests/test_modules_lora.py
tests/test_moe_experts.py
tests/test_mrope_widen.py
tests/test_mtp.py
tests/test_mtp_preempt_resume.py
tests/test_mtp_width_cap.py
tests/test_muse_glimmer_mtp.py
tests/test_muse_glimmer_tools.py
tests/test_muse_glimmer_vision.py
tests/test_native_fp.py
tests/test_occupancy_fuse.py
tests/test_offload.py
tests/test_overgen.py
tests/test_pagecache.py
tests/test_patching.py
tests/test_pin_weights.py
tests/test_populate.py
tests/test_pq_log.py
tests/test_prefetch.py
tests/test_prefill_decay.py
tests/test_prefill_ticks.py
tests/test_prefix_cache_budget.py
tests/test_preflight.py
tests/test_profiles.py
tests/test_quantized_sdpa_fix.py
tests/test_qwen35_attn.py
tests/test_qwen35_gdn.py
tests/test_qwen35_layers.py
tests/test_qwen35_mirrors.py
tests/test_qwen35_owned.py
tests/test_qwen35_rope.py
tests/test_qwen35_verify_fold.py
tests/test_qwen35_verify_linear.py
tests/test_ragged_decode.py
tests/test_reasoning.py
tests/test_remap.py
tests/test_remote.py
tests/test_render.py
tests/test_rerank.py
tests/test_residency.py
tests/test_residency_config.py
tests/test_retire_key.py
tests/test_rope_batch_fix.py
tests/test_rope_factors.py
tests/test_rotating_inject_extend.py
tests/test_rotating_snapshot.py
tests/test_run_think_echo.py
tests/test_serve_apc_engagement.py
tests/test_serve_memtrace.py
tests/test_server.py
tests/test_server_bridge.py
tests/test_server_memory.py
tests/test_server_patches.py
tests/test_serving_apc_config.py
tests/test_serving_config.py
tests/test_serving_mtp.py
tests/test_serving_text_embed.py
tests/test_serving_vlm.py
tests/test_shared_kv_inject_merge.py
tests/test_sparse_sdpa.py
tests/test_spec_engine_release.py
tests/test_spec_kv_quant.py
tests/test_spinner.py
tests/test_stoch_accept.py
tests/test_structured.py
tests/test_stt.py
tests/test_switch_compat.py
tests/test_talk.py
tests/test_talk_audio.py
tests/test_talk_client.py
tests/test_talk_mcp.py
tests/test_talk_memory.py
tests/test_theme.py
tests/test_thinking_budget.py
tests/test_tokenizer.py
tests/test_train.py
tests/test_transforms.py
tests/test_tts.py
tests/test_upstream_seams.py
tests/test_verify_ragged_mask.py
tests/test_vlm_integration.py
tests/test_vlm_mtp_gating.py
tests/test_vlm_mtp_integration.py
tests/test_vlm_resolve.py
tests/test_wire_contract.py
tests/test_wired_limit_patch.py
tests/e2e/README.md
tests/e2e/checks.py
tests/e2e/client.py
tests/e2e/judge.py
tests/e2e/models.py
tests/e2e/prompts.py
tests/e2e/pty_session.py
tests/e2e/report.py
tests/e2e/run_apc_depth_e2e.py
tests/e2e/run_apc_disk_e2e.py
tests/e2e/run_chat_pty_e2e.py
tests/e2e/run_lora_e2e.py
tests/e2e/run_serve_harmony_e2e.py
tests/e2e/run_server_e2e.py
tests/e2e/scenarios.py
tests/e2e/server_proc.py
tests/e2e/assets/README.md
tests/e2e/assets/cats.jpg
tests/e2e/assets/speech.wav
tests/fixtures/kimi_k3_template.jinja