{"uid":"cap_kBIGrb9qVxQYmSuau5JUO","slug":"salt19-model-memory-fit-estimator-17e02678","name":"SALT19 Model Memory Fit Estimator","description":"Machine-discoverable, x402-paid technical decision utilities for autonomous agents.","url":"https://salt19-agent-utility-grid.salt19-llc.workers.dev/v1/model-memory-fit","method":"POST","headers":{},"bodySchema":{"type":"object","properties":{"vram_gb":{"type":"number","maximum":10000,"minimum":0.1},"parameter_count_b":{"type":"number","maximum":10000,"minimum":0.01},"quantization_bits":{"type":"number","maximum":32,"minimum":1},"runtime_overhead_ratio":{"type":"number","maximum":2,"minimum":0}}},"responseSchema":{"type":"json","example":{"verdict":"FIT","vram_gib":11,"limitations":["KV cache and architecture-specific buffers are not modeled"],"headroom_gib":7.25,"estimated_total_gib":3.75,"estimated_weight_gib":3.26}},"example":null,"exampleRequest":null,"tags":["x402"],"displayCostAmount":"0.05","displayCostAsset":"USDC","priceDynamic":false,"priceHint":null,"priceStatus":"priced","priceSource":"probe","requiresHandshake":false,"reviewCount":0,"rating":{"score":"0.00","successRate":"0.00","reviews":0,"stars":null,"state":"unrated"},"availabilityStatus":"unknown","priceObserved":null,"sessionDeposit":null,"pricing":{"kind":"static","summary":"$0.05/call","primary":{"kind":"static","protocol":"x402","network":"base","amountUsd":"0.05","per":"call","confidence":"exact"},"accepted":[{"kind":"static","protocol":"x402","network":"base","amountUsd":"0.05","per":"call","confidence":"exact"}]},"paymentMethods":[{"uid":"pm_5bejbqaas5cuhyr2K8KJ7","protocol":"x402","methodType":"crypto","chain":"base","mode":"charge","costAmount":"0.05","costPer":"request","priority":0,"asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913","unit":"request","depositMicros":null,"planRef":null}],"brandName":null,"brandSlug":null,"brandBaseUrl":null,"brandDocsUrl":null,"whatItDoes":"Estimates whether a language model with given parameter count and quantization will fit in a specified VRAM budget, returning a FIT/NO-FIT verdict with memory breakdown","exampleAgentPrompt":"Will a 7 billion parameter model fit on a GPU with 16 GB of VRAM if I quantize it to 4 bits, assuming a 10% runtime overhead?","exampleUseCases":[{"title":"Consumer GPU deployment check","prompt":"I have an RTX 3080 with 10 GB of VRAM and I want to run a 13B parameter model at 4-bit quantization with about 5% overhead — will it actually fit?"},{"title":"Data center multi-model planning","prompt":"I'm planning to load a 70 billion parameter model on an A100 with 80 GB of VRAM using 8-bit quantization and 20% runtime overhead — can you tell me if it fits and how much headroom I'll have?"},{"title":"Optimal quantization selection","prompt":"I only have 24 GB of VRAM and want to run a 34 billion parameter model — check if it fits at 4 bits with a 15% overhead ratio so I know if I need to quantize further."}],"resultDescription":"A JSON object containing a 'verdict' field ('FIT' or similar), 'estimated_weight_gib' (model weights memory), 'estimated_total_gib' (total estimated memory usage), 'headroom_gib' (remaining VRAM after model load), 'vram_gib' (the input VRAM in GiB), and a 'limitations' array noting what is not modeled (e.g. KV cache, architecture-specific buffers).","failureModes":["Invalid vram_gb outside 0.1–10000 range returns validation error","Invalid parameter_count_b outside 0.01–10000 range returns validation error","quantization_bits outside 1–32 returns validation error","Missing required fields may return 400 or default behavior","Payment failure via x402 protocol returns 402 Payment Required","Network timeout on Cloudflare Workers edge"],"whenToPreferThis":"Choose this endpoint when an agent needs a quick, programmatic, paid-per-call estimate of whether a specific LLM configuration will fit into a GPU's VRAM budget before attempting to load the model. Ideal for automated deployment pipelines, agent-driven infrastructure decisions, or when comparing quantization strategies across multiple GPU tiers. Prefer this over manual calculation when parameter count, quantization, and overhead ratio are all known and a structured FIT/NO-FIT verdict with memory breakdown is needed.","instructions":null,"reviewSummary":null,"reviewSummaryHighlights":null,"reviewSummaryConcerns":null,"reviewSummaryGeneratedAt":null,"activationCount":0,"lastUsedAt":null,"lastSuccessfullyRanAt":null,"lastHealthCheckAt":"2026-09-15T17:21:49.255Z","isFirstParty":false}