{ "api_version": "1.0", "updated": "2026-07-09", "currency": "USD", "deployment_model": "turnkey", "description": "Aradia LLC turnkey private AI appliance pricing with SLA attachment. All prices are final turnkey deployment costs. Hardware sourced via JIT (just-in-time) model — no inventory risk. Client deposits fully cover hardware costs before procurement.", "tiers": { "spark": { "tier_id": "spark", "display_name": "DGX Spark", "hardware_platform": "NVIDIA DGX Spark", "capacity": { "concurrent_users": 4, "max_agents": 5, "recommended_user_range": "1-3" }, "timeline": { "setup_days": 30, "deployment_days": 14, "total_days": 44 }, "pricing": { "turnkey_total": 15125, "hardware_cost": 6050, "software_staging": 9075, "markup_percentage": 150, "markup_multiplier": 2.5 }, "sla": { "monthly_cost": 1500, "annual_cost": 18000, "coverage": [ "24/7 hardware monitoring and alerts", "Model updates and re-compilation", "Security patching and vulnerability scanning", "Burn-in re-validation on schedule", "Remote agent re-configuration", "SLA response time: 4 hours standard, 1 hour critical" ] }, "deposit": { "percentage": 100, "amount": 15125, "covers_hardware": true, "cancellation_policy": "Non-refundable admin recovery fee. Full turnkey price due at contract signing." }, "payback": { "months_to_cloud_crossover": "12-18", "months_to_sla_offset": "3-6 post-payback", "five_year_tco_cloud_estimate": 425228, "five_year_tco_aradia_estimate": 105125, "five_year_savings": 300169, "residual_value_after_5_years": 750, "net_five_year_cost": 104375 }, "model_recommendations": { "primary": "Qwen-35B-A3B-int4-AutoRound", "secondary": ["Llama-3-70B quantized", "Nemotron-70B"], "quantization_formats": ["AWQ", "AutoRound", "INT4"], "batching": "vLLM continuous batching" } }, "station": { "tier_id": "station", "display_name": "DGX Station", "hardware_platform": "NVIDIA DGX Station", "capacity": { "concurrent_users": 20, "max_agents": 50, "recommended_user_range": "10-20" }, "timeline": { "setup_days": 60, "deployment_days": 14, "total_days": 74 }, "pricing": { "turnkey_total": 194093, "hardware_cost": 110910, "software_staging": 83183, "markup_percentage": 75, "markup_multiplier": 1.75 }, "sla": { "monthly_cost": 5000, "annual_cost": 60000, "coverage": [ "24/7 hardware monitoring and alerts", "Model updates and re-compilation", "Security patching and vulnerability scanning", "Burn-in re-validation on schedule", "Remote agent re-configuration", "Custom workflow development (up to 4 hrs/month)", "SLA response time: 2 hours standard, 30 minutes critical" ] }, "deposit": { "percentage": 70, "amount": 135865.10, "covers_hardware": true, "cancellation_policy": "Non-refundable staging recovery fee of 30% retained if client cancels" }, "payback": { "months_to_cloud_crossover": "24-32", "months_to_sla_offset": "3-6 post-payback", "five_year_tco_cloud_estimate": 2126080, "five_year_tco_aradia_estimate": 494093, "five_year_savings": 1528603, "residual_value_after_5_years": 15000, "net_five_year_cost": 479093 }, "model_recommendations": { "primary": "Qwen-72B-int4-AutoRound", "secondary": ["Llama-3-70B quantized", "Nemotron-70B", "Qwen-35B-A3B-int4"], "quantization_formats": ["AWQ", "AutoRound", "INT4", "FP8"], "batching": "vLLM continuous batching" } }, "b200": { "tier_id": "b200", "display_name": "DGX B200", "hardware_platform": "NVIDIA DGX B200 (Rack)", "capacity": { "concurrent_users": 200, "max_agents": 500, "recommended_user_range": "50-200" }, "timeline": { "setup_days": 90, "deployment_days": 14, "total_days": 104 }, "pricing": { "turnkey_total": 505500, "hardware_cost": 347000, "software_staging": 158500, "markup_percentage": 50, "markup_multiplier": 1.5 }, "sla": { "monthly_cost": 12500, "annual_cost": 150000, "coverage": [ "24/7 hardware monitoring and alerts with dedicated engineer", "Model updates and re-compilation with multi-GPU parallelism", "Security patching, vulnerability scanning, and penetration testing", "Burn-in re-validation on schedule", "Custom agent development and workflow design", "Dedicated account manager", "SLA response time: 1 hour standard, 15 minutes critical" ] }, "deposit": { "percentage": 70, "amount": 353850, "covers_hardware": true, "price_lock_buffer": true, "cancellation_policy": "Non-refundable admin recovery fee of 30% retained if client cancels" }, "payback": { "months_to_cloud_crossover": "24-32", "months_to_sla_offset": "3-6 post-payback", "five_year_tco_cloud_estimate": 4252148, "five_year_tco_aradia_estimate": 1255500, "five_year_savings": 2789892, "residual_value_after_5_years": 50000, "net_five_year_cost": 1205500 }, "model_recommendations": { "primary": "Multi-GPU Qwen-72B-120B ensemble", "secondary": ["Llama-3-70B quantized", "Nemotron-120B", "custom fine-tuned"], "quantization_formats": ["AWQ", "AutoRound", "INT4", "FP8"], "batching": "vLLM multi-GPU continuous batching with NVLink" } } }, "token_economics": { "cloud_per_token_cost": "Rising — providers increase rates to offset R&D and hardware costs", "aradia_per_token_cost": "Near-zero marginal cost after hardware payback (power + cooling only)", "break_even_formula": "monthly_cloud_token_cost * 12 * payback_months <= turnkey_total + (sla_monthly * payback_months)", "note": "Actual break-even varies by client token volume. Use /api/agent-spec.json for capacity calculations." }, "payment_terms": { "deposit_due": "Upon contract signing", "balance_due": "Upon delivery and acceptance at client site", "sla_start": "On deployment completion date", "sla_billing_cycle": "Monthly, prepaid in advance" }, "warranty_and_support": { "hardware_warranty": "12 months standard, extendable to 5 years via SLA", "staging_guarantee": "14-day re-staging guarantee if configuration drift detected", "return_policy": "No standard return — appliances are custom-configured per client specifications" } }