{ "title": "Nova Nano 1.0 performance lab", "status": "All numerical performance values are theoretical or not validated; no unverified benchmark is presented as measured", "status legend": { "MEDIDO": "physical measurement backed by a dated log and exact configuration", "ESTIMADO": "mathematical projection with its assumptions documented", "NO VALIDADO": "insufficient evidence or unavailable native configuration" }, "repository": "ModernoTech/Nova-Nano", "artifact": { "filename": "Nova-Nano-1.0-Q4_K_M.gguf", "bytes": 229312224, "sha256": "7e6f72643caafc9a68256686638c4d7916f2cec76d1df478d4c3ddcd95a6aed4", "format": "GGUF", "published": true, "source_of_size": "Hugging Face file metadata" }, "specs": { "model_name": "Nova Nano 1.0", "base_reference": "LiquidAI/LFM2.5-350M", "total_params_billion": 0.35, "active_params_billion": 0.35, "moe": false, "max_position_documented": 128000 }, "architecture_basis": { "layers": 16, "kv_heads": 8, "head_dim": 128, "source": "LiquidAI config.json downloaded from the documented base reference", "status": "ESTIMADO sobre una arquitectura publicada de terceros" }, "estimation_assumptions": { "ram_bandwidth_values_gbps": { "MacBook Air 2014": 25.6, "Lenovo Yoga 720": 34.1, "MacBook Neo A18 Pro": 60.0, "MacBook Air M4": 120.0 }, "ssd_effective_gbps": 2.0, "ssd_basis": "Assumed external NVMe 2.0 GB/s for calculations only; must be replaced by a measured sequential/shared load if used externally", "kv_precision_bytes": 2.0, "weight_precision_bytes": 0.5, "overhead_factor": 1.1, "os_overhead_gb": 1.2, "runtime_overhead_bytes": 120.0, "latency_included": "Fixed read latency excluded; remeasurement is required before any availability claim", "mne_expert_cache_fraction": "Active set + capped additive cache; rounded, saturation assumptions stated", "conventional_write_notes": "Conventional RAM path counts total artifact weights resident in RAM; active path measures only the bytes read per token as an upper bound" }, "measured": null, "ssd_cache_scenarios": [ { "cache_hit": 0.9, "ssd_mne_upper_tok_s": 103.896 }, { "cache_hit": 0.95, "ssd_mne_upper_tok_s": 207.792 }, { "cache_hit": 0.99, "ssd_mne_upper_tok_s": 1038.961 }, { "cache_hit": 0.995, "ssd_mne_upper_tok_s": 2077.922 }, { "cache_hit": 0.999, "ssd_mne_upper_tok_s": 10389.61 } ], "scenarios": [ { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Air 2014", "hardware_id": "air-2014", "bandwidth_gbps": 25.6, "ram_gb": 4.0, "context_tokens": 4096, "context_label": "4K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 132.987, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 268435456, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.580751, "mne_resident_total_gb_est": 0.908438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "Lenovo Yoga 720", "hardware_id": "yoga-720", "bandwidth_gbps": 34.1, "ram_gb": 8.0, "context_tokens": 4096, "context_label": "4K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 177.143, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 268435456, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.580751, "mne_resident_total_gb_est": 0.908438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Neo A18 Pro", "hardware_id": "neo", "bandwidth_gbps": 60.0, "ram_gb": 8.0, "context_tokens": 4096, "context_label": "4K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 311.688, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 268435456, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.580751, "mne_resident_total_gb_est": 0.908438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Air M4", "hardware_id": "air-m4", "bandwidth_gbps": 120.0, "ram_gb": 16.0, "context_tokens": 4096, "context_label": "4K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 623.377, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 268435456, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.580751, "mne_resident_total_gb_est": 0.908438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Air 2014", "hardware_id": "air-2014", "bandwidth_gbps": 25.6, "ram_gb": 4.0, "context_tokens": 8192, "context_label": "8K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 132.987, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 536870912, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.830751, "mne_resident_total_gb_est": 1.158438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "Lenovo Yoga 720", "hardware_id": "yoga-720", "bandwidth_gbps": 34.1, "ram_gb": 8.0, "context_tokens": 8192, "context_label": "8K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 177.143, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 536870912, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.830751, "mne_resident_total_gb_est": 1.158438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Neo A18 Pro", "hardware_id": "neo", "bandwidth_gbps": 60.0, "ram_gb": 8.0, "context_tokens": 8192, "context_label": "8K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 311.688, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 536870912, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.830751, "mne_resident_total_gb_est": 1.158438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." }, { "model": "Nova Nano 1.0", "repository": "ModernoTech/Nova-Nano", "hardware": "MacBook Air M4", "hardware_id": "air-m4", "bandwidth_gbps": 120.0, "ram_gb": 16.0, "context_tokens": 8192, "context_label": "8K", "context_supported_or_validated": true, "max_position_documented": 128000, "artifact_size_bytes": 229312224, "artifact_status": "PUBLICADO", "weight_basis": "artifact size on Hugging Face", "active_payload_bytes_for_memory_bound": 206695301, "active_payload_basis": "active param count x 0.5 GiB/B x 1.10 (Q4 + 10% parsed overhead), ESTIMADO", "theoretical_tok_s_conventional_ram": 623.377, "ssd_mne_upper_tok_s": [ { "ssd_cache_miss": 0.1, "cache_label": "90%", "upper Tok_s": 103.896 }, { "ssd_cache_miss": 0.05, "cache_label": "95%", "upper Tok_s": 207.792 }, { "ssd_cache_miss": 0.01, "cache_label": "99%", "upper Tok_s": 1038.961 }, { "ssd_cache_miss": 0.005, "cache_label": "99.5%", "upper Tok_s": 2077.922 }, { "ssd_cache_miss": 0.001, "cache_label": "99.9%", "upper Tok_s": 10389.61 } ], "kv_cache_bytes_est": 536870912, "kv_cache_status": "ESTIMADO", "conventional_resident_total_gb_est": 0.830751, "mne_resident_total_gb_est": 1.158438, "ssd_required_bytes": 229312224, "feasible_conventional": "VIABLE con margen de memoria (límite superior de presión, no medición)", "feasible_mne": "ESCENARIO VIRTUOSO DE RAM (cache capado; no prueba de throughput)", "throughput_status": "ESTIMADO (techo de ancho de banda, no benchmark)", "throughput_note": "No real benchmark is stored for MNE; upper bounds only." } ], "charts": [ "tokens-per-second.png", "ram-vs-mne.png", "context-performance.png", "memory-breakdown.png", "model-size-vs-ram.png", "cache-hit-analysis.png", "prefill-vs-decode.png" ] }