{ "arch_id": "qwen3-next-mtp", "artifact_role": "qwen35-9b-optimized-speed-fp16", "base_trunk": "Qwen/Qwen3.5-9B", "exactness_baseline": {}, "forge_provenance": { "local_paths_required": false, "release_repo": "Youssofal/Qwen3.5-9B-MTPLX-Optimized-Speed-FP16", "source_artifact": "Qwen3.5-9B MTPLX Optimized Speed" }, "hub": { "published_name": "Qwen3.5-9B-MTPLX-Optimized-Speed-FP16", "repo_id": "Youssofal/Qwen3.5-9B-MTPLX-Optimized-Speed-FP16" }, "mtp_contract": { "base_hidden_variant": "post_norm", "concat_order": "embedding_hidden", "hidden_variant": "post_norm", "mtp_position_mode": "local", "mtp_quant_group_size": 64, "mtp_quant_mode": "affine" }, "mtp_depth_max": 3, "mtp_sidecar": "mtp.safetensors", "mtplx_version": "0.3.8", "precision_policy": { "intended_default_for": [ "m1", "m2" ], "note": "Sibling compatibility variant; not a universal speed claim.", "policy": "convert_bf16_float_tensors_to_fp16_preserve_packed_quantized_tensors", "preserve_mtp_compatibility": true, "routing": "MTPLX app and CLI prefer this artifact on M1/M2 Apple Silicon when the model is available.", "source_repo": "Youssofal/Qwen3.5-9B-MTPLX-Optimized-Speed", "variant": "fp16" }, "precision_variant": "fp16", "promoted_decode_depth": 2, "public_model_id": "mtplx-qwen35-9b-optimized-speed-fp16", "recommended_mtp_depth": 2, "recommended_profile": "sustained", "routing": { "catalog_preference": "legacy_apple_fp16", "intended_chip_tiers": [ "m1", "m2" ] }, "sampler": { "temperature": 0.6, "top_k": 20, "top_p": 0.95 }, "served_model_id": "mtplx-qwen35-9b-optimized-speed-fp16", "speed_evidence": { "acceptance_by_depth": [ 0.9398148148148148, 0.8101851851851852 ], "acceptance_collapsed": [], "depth": 2, "failure_reasons": [], "forge_verify_rows": [ { "acceptance_by_position": [], "depth": 0, "finish_reasons": { "stop": 1 }, "hit_token_budget": false, "hit_token_budget_count": 0, "multiplier_vs_ar": 1.0, "quality_passed": true, "tok_s": 64.96364640251394, "verify_time_s": 13.320888499962166 }, { "acceptance_by_position": [ 0.9120370370370371 ], "depth": 1, "finish_reasons": { "stop": 1 }, "hit_token_budget": false, "hit_token_budget_count": 0, "multiplier_vs_ar": 1.4295825302162772, "quality_passed": true, "tok_s": 92.87089399618144, "verify_time_s": 6.827253244584426 }, { "acceptance_by_position": [ 0.9398148148148148, 0.8101851851851852 ], "depth": 2, "finish_reasons": { "stop": 1 }, "hit_token_budget": false, "hit_token_budget_count": 0, "multiplier_vs_ar": 1.5596130963419594, "quality_passed": true, "tok_s": 101.31815371548896, "verify_time_s": 4.125403580837883 }, { "acceptance_by_position": [ 0.9278350515463918, 0.7731958762886598, 0.6443298969072165 ], "depth": 3, "finish_reasons": { "stop": 1 }, "hit_token_budget": false, "hit_token_budget_count": 0, "multiplier_vs_ar": 1.4823467156220833, "quality_passed": true, "tok_s": 96.29864787960092, "verify_time_s": 4.5755067388527095 } ], "greedy_diagnostic": { "tok_s": 64.96364640251394 }, "quality_rejected": [], "tok_s": [ 101.31815371548896 ], "verdict": "mtp_depth_wins" }, "verified_on": { "hardware": "macOS-26.3.1-arm64-arm-64bit-Mach-O", "machine_arch": "arm64", "macos": "26.3.1", "model": "Qwen3.5-9B MTPLX Optimized Speed", "timestamp": "2026-06-06T06:20:32+01:00" } }