{"name":"PROXIES.SX Compute","description":"Early-access managed inference rentals on provider-owned Apple Silicon Macs.","url":"https://www.proxies.sx/compute","reviewedAt":"2026-09-11","updatedAt":"2026-09-11","reviewScope":"Public documentation, catalog, tiers, fleet and marketplace responses. No authenticated inference or payout test.","snapshot":{"date":"2026-09-11","listedMachines":0,"onlineNodes":0,"catalogEntries":3,"publishedTiers":3,"live":false},"billing":"Published tiers are per 30 days. Confirm current price and stock in the product API and dashboard.","providerShare":{"approval":"Per-node admin terms","defaultPercentSnapshot":45,"snapshotDate":"2026-09-11","basis":"Agreed share of paid rentals only; free allocations earn nothing.","settlement":"Accrual, settlement and completed payout are separate states.","source":"https://agents.proxies.sx/compute/facts.json"},"pricing":{"authority":"admin","nodeOverrides":true,"quoteSource":"https://api.proxies.sx/v1/peer/compute/marketplace","quoteFields":["expectedPriceUsd","pricingVersion"],"termDays":30,"autoRenew":false,"perTokenBilling":false,"checkout":"https://compute.proxies.sx/market"},"catalogSnapshot":{"reviewedAt":"2026-09-11","models":[{"id":"qwen3.8-27b-4bit","hfRepo":"mlx-community/Qwen3.8-27B-4bit","revision":"3e6447f082e89cc7f0bc6e5441afd38dfce760ff","family":"Qwen3.8-27B","quant":"4bit","maxContext":16384,"requiredMemGB":24,"approxDiskGB":16},{"id":"qwen3.8-27b-8bit","hfRepo":"mlx-community/Qwen3.8-27B-8bit","revision":"815b83c0df8ffd1d1b5244cf75fd6ef14fca9ef9","family":"Qwen3.8-27B","quant":"8bit","maxContext":32768,"requiredMemGB":48,"approxDiskGB":30},{"id":"qwen3.8-27b-bf16","hfRepo":"mlx-community/Qwen3.8-27B-bf16","revision":"6f265714824f3c38d4452baa1628aef3d9b9aae9","family":"Qwen3.8-27B","quant":"bf16","maxContext":32768,"requiredMemGB":96,"approxDiskGB":55}]},"productFacts":"https://agents.proxies.sx/compute/facts.json","discovery":[{"name":"catalog","url":"https://api.proxies.sx/v1/peer/compute/catalog","method":"GET","authentication":"none","purpose":"read-only discovery"},{"name":"tiers","url":"https://api.proxies.sx/v1/peer/compute/tiers","method":"GET","authentication":"none","purpose":"read-only discovery"},{"name":"marketplace","url":"https://api.proxies.sx/v1/peer/compute/marketplace","method":"GET","authentication":"none","purpose":"read-only discovery"},{"name":"stats","url":"https://api.proxies.sx/v1/peer/compute/stats","method":"GET","authentication":"none","purpose":"read-only discovery"}],"customerLogin":"https://compute.proxies.sx/login","providerDocumentation":"https://agents.proxies.sx/compute/","providerOnboarding":{"accountRequired":true,"signup":"https://farmer.proxies.sx/signup","emailVerification":"https://farmer.proxies.sx/verify-email","agreement":"https://farmer.proxies.sx/agreement","keyScopes":["account:read","compute:supply"],"supplierApprovalRequired":true,"dashboard":"https://farmer.proxies.sx/compute","earnings":"https://farmer.proxies.sx/earnings","payouts":"https://farmer.proxies.sx/payouts","steps":["Create a farmer account and verify your email","Sign the Partner Agreement and create a compute supply key in the farmer Compute page","Install the compute agent on an eligible Mac with that key","Complete model readiness, the benchmark and supplier approval, review approved price and share, then list the node in the farmer compute dashboard","Manage earnings and payout requests in the farmer portal"],"reviewedAt":"2026-09-11","verificationScope":"Public signup client, portal routes and provider documentation; no account registration or payout performed."},"inference":{"authentication":"Active rental and account credentials; documented X-API-Key header.","documentation":"https://www.proxies.sx/compute/api","keyScopes":["compute:read","compute:infer"],"unsupportedRequestFields":["tools","stop","response_format"],"maxOutputTokens":4096,"requestsPerMinutePerRental":60,"contractBasis":"Published customer documentation; not an authenticated API test.","x402ComputePurchaseVerified":false},"text":"https://www.proxies.sx/compute/llms.txt","evidenceSnapshot":"https://www.proxies.sx/compute/review.json","pages":[{"url":"https://www.proxies.sx/compute/rent","title":"Rent an Apple Silicon Mac for dedicated inference","summary":"A dedicated endpoint makes sense when you have recurring model traffic and want to reserve a machine for it. Start with the model your application needs, then check whether a listed Mac can serve it.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"},{"title":"Marketplace inventory API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/marketplace"},{"title":"Compute fleet statistics API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/stats"}],"description":"Evaluate a dedicated Apple Silicon inference rental: 30-day terms, supported models, API limits, stock and the checks to make before paying.","audience":"Renters","evidence":{"title":"What happens between finding a Mac and running a prompt?","answer":"A balance top-up does not reserve compute. A rental exists after the quoted purchase successfully allocates a machine. Renew manually at a new quote for another 30 days.","facts":[{"value":"30 days","label":"Paid term"},{"value":"Manual","label":"Renewal"},{"value":"No token fee","label":"Request limits still apply"}],"visual":"steps","steps":[{"title":"Check ready stock","detail":"Find a suitable machine and review its effective price before adding funds."},{"title":"Sign in and fund balance","detail":"Use Stripe card checkout or the CoinGate options offered at checkout."},{"title":"Confirm the current quote","detail":"The portal checks price and pricing version. A pending operation is unresolved."},{"title":"Use the allocated endpoint","detail":"A completed purchase supplies the rental. Confirm model readiness, then send text requests."}],"caption":"Documented purchase sequence, not a completed checkout test. Failed allocation reverses its charge to account balance.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Understand price changes and rental terms","path":"/compute/pricing"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/providers","title":"Rent out your Mac for AI inference","summary":"A farmer account is required to provide compute. Create it first, then connect a Mac you already own and can keep available. Your farmer portal is where you manage nodes, listings, earnings and payout requests.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Compute agent source","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/compute-agent.js"},{"title":"Python virtual environments","publisher":"Python Software Foundation","url":"https://docs.python.org/3/library/venv.html"},{"title":"Node.js release schedule","publisher":"Node.js project","url":"https://nodejs.org/en/about/previous-releases"}],"description":"Create your required farmer account, verify email and install the Mac compute agent. Follow benchmark and supplier approval through listing and earnings.","audience":"Providers","evidence":{"title":"An installed agent is one step toward an approved listing.","answer":"You need a verified farmer account, a countersigned agreement, a ready model, a passing benchmark and admin approval. Review the approved price and share in the Compute dashboard.","facts":[{"value":"0.3.1","label":"Reviewed agent version"},{"value":"24 GB","label":"Minimum installed RAM"},{"value":"26 GB","label":"Free disk for smallest setup"}],"visual":"steps","steps":[{"title":"Account and agreement","detail":"Create your farmer account, verify email and complete the signed and countersigned agreement."},{"title":"Supply key and runtime","detail":"Create the key in Farmer → Compute. Install the native Apple Silicon agent on macOS 14+."},{"title":"Readiness and review","detail":"The pinned model must serve successfully. Benchmark verification and admin admission are separate checks."},{"title":"Approved terms and listing","detail":"Check locked pricing and your share. A proposed price has no effect until admin approves it."}],"caption":"Keep the Mac plugged in, awake and online. Starting at login does not prevent lid-close sleep. Listing and free tests do not earn rental income.","sources":[{"label":"Provider documentation","url":"https://agents.proxies.sx/compute/"},{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Estimate earnings with your approved share","path":"/compute/provider-earnings"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/apple-silicon","title":"Apple Silicon memory and performance for LLM inference","summary":"The chip name alone cannot tell you which model a Mac will serve. Start with usable memory, account for the context your workload needs, then measure performance on that configuration.","references":[{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Model catalog API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"title":"MLX unified memory","publisher":"MLX project","url":"https://ml-explore.github.io/mlx/build/html/usage/unified_memory.html"},{"title":"View memory usage in Activity Monitor on Mac","publisher":"Apple","url":"https://support.apple.com/guide/activity-monitor/view-memory-usage-actmntr1004/mac"}],"description":"Check Mac memory for LLM inference. Understand weights, KV cache, memory pressure, disk space and the current 24 GB, 48 GB and 96 GB catalog requirements.","audience":"Providers","evidence":{"title":"Memory requirements are not download sizes.","answer":"The smallest catalog configuration requires 24 GB of RAM and approximately 16 GB on disk. Model weights, runtime allocations, context state and macOS use different parts of the capacity budget.","facts":[{"value":"24 / 48 / 96 GB","label":"Catalog memory requirements"},{"value":"16 / 30 / 55 GB","label":"Approximate model disk sizes"},{"value":"Separate review","label":"Verified tier and admission"}],"visual":"catalog","caption":"Switch the chart metric to inspect RAM, disk or context separately. Installed RAM alone does not establish a tier or guarantee rental demand.","sources":[{"label":"Live model catalog","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"label":"Provider documentation","url":"https://agents.proxies.sx/compute/"}],"next":{"label":"Measure memory pressure and response time","path":"/compute/benchmarks"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/models","title":"Compute model catalog and memory requirements","summary":"The public catalog identifies models and their memory and context requirements. Match a pinned model revision to an approved machine listing before planning a rental.","references":[{"title":"Compute agent source","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/compute-agent.js"},{"title":"Model catalog API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"title":"Marketplace inventory API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/marketplace"},{"title":"Download files from the Hub","publisher":"Hugging Face","url":"https://huggingface.co/docs/huggingface_hub/guides/download"},{"title":"Model cards","publisher":"Hugging Face","url":"https://huggingface.co/docs/hub/model-cards"}],"description":"Compare Qwen3.8 27B model IDs, pinned revisions, quantization, RAM, disk and context limits. Check the reviewed catalog against current rental stock.","audience":"Renters","evidence":{"title":"Compare the supported catalog by the limit you need.","answer":"All three entries belong to Qwen3.8 27B. Precision, required memory, disk and context differ. The exact model IDs and immutable revisions appear in the tables below.","facts":[{"value":"3 entries","label":"In the reviewed catalog"},{"value":"27B","label":"Model family size"},{"value":"Pinned","label":"Repository revisions"}],"visual":"catalog","caption":"Data from the September 11 catalog response. Model support is separate from rentable stock; confirm both before choosing a machine.","sources":[{"label":"Live model catalog","url":"https://api.proxies.sx/v1/peer/compute/catalog"}],"next":{"label":"Evaluate 4-bit, 8-bit and BF16 quality","path":"/compute/quantization"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/pricing","title":"Dedicated inference pricing for a 30-day Mac rental","summary":"Consider a dedicated endpoint when its model passes your evaluation and you have recurring work for it. Compare the full 30-day rental price with the token bill for that workload.","references":[{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"},{"title":"Rental tiers API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/tiers"},{"title":"Marketplace inventory API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/marketplace"}],"description":"Compare dated Mac rental defaults with approved machine quotes. Understand 30-day balance payments, manual renewal, request limits and useful-output costs.","audience":"Renters","evidence":{"title":"A tier default is a reference, not your machine quote.","answer":"Admin may approve a different price for an individual Mac. Checkout uses the effective machine price and pricing version. Accepted terms remain fixed for that intent; a new renewal needs a current quote.","facts":[{"value":"USD balance","label":"Compute purchase funding"},{"value":"30 days","label":"Term, not a calendar month"},{"value":"Per node","label":"Approved price overrides"}],"visual":"prices","caption":"Admin defaults returned September 11, 2026, in USD per 30 days. These bars are neither current offers nor an availability claim.","sources":[{"label":"Live tier defaults","url":"https://api.proxies.sx/v1/peer/compute/tiers"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Follow the rental purchase sequence","path":"/compute/rent"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/api","title":"Compute API integration for AI agents","summary":"An agent can read the public catalog before it has an account. Sending inference requests requires an active rental and its credentials. Keep discovery, purchasing and model execution as separate application steps.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"}],"description":"Integrate dedicated Mac inference with scoped API keys. Review request limits, streaming, unsupported options, errors and safe discovery before checkout.","audience":"Developers","evidence":{"title":"Frequency, queue capacity and generation time are different limits.","answer":"The published limit is 60 requests per minute per rental, with one active job per node. A long answer can occupy that node while other requests wait. Measure completed work as well as admitted requests.","facts":[{"value":"4,096 tokens","label":"Maximum output per request"},{"value":"60 / minute","label":"Request rate per rental"},{"value":"1 job","label":"Active on a node"}],"visual":"steps","steps":[{"title":"Validate the request","detail":"Use the assigned model, text messages and an output budget. Tools and response_format are unsupported."},{"title":"Wait for the node","detail":"Bound your application queue. The request-rate allowance does not promise immediate execution."},{"title":"Generate and stream","detail":"The documented generation deadline is 180 seconds; idle generation fails after 60 seconds."},{"title":"Handle the result","detail":"Check completion and errors. Use a deadline and a bounded retry policy for inference."}],"caption":"A schematic of documented limits, not a measured latency timeline. Use a read/inference key for model requests and keep purchase permissions separate.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Record latency and completed throughput","path":"/compute/benchmarks"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/provider-earnings","title":"Mac compute provider earnings after operating costs","summary":"The useful number is the amount left after costs and unrented time. Use the calculator with a machine you own, a measured power figure and an occupancy assumption you are prepared to question.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Marketplace inventory API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/marketplace"},{"title":"Compute fleet statistics API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/stats"}],"description":"Estimate Mac provider earnings with your approved rental price and share. Compare occupancy, electricity and costs; separate accrual, settlement and payouts.","audience":"Providers","evidence":{"title":"Earned share, settled balance and paid cash are different amounts.","answer":"For each successful paid term, accrual follows its accepted price and approved provider share. Free test allocations earn nothing. Use the rental ledger to check earned amounts and the payout record to check received funds.","facts":[{"value":"Editable","label":"Approved share and rental price"},{"value":"Paid terms","label":"Basis for actual accrual"},{"value":"No guarantee","label":"Occupancy assumptions"}],"visual":"steps","steps":[{"title":"Paid rental accrues your share","detail":"Use the accepted price × paid terms × agreed share. Sum terms separately if their pricing differs."},{"title":"Review and settle earnings","detail":"The dated snapshot reports shadow settlement mode. Admin review and settlement are required."},{"title":"Track the completed payout","detail":"A payout request is not paid cash. Check its final state in the farmer portal."}],"caption":"Existing peer ledger entries are already net of the platform share. Do not subtract that percentage a second time.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Review provider admission and pricing","path":"/compute/providers"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/benchmarks","title":"Benchmark Apple Silicon inference on your workload","summary":"A tokens-per-second number is incomplete without the model, prompt length and measurement method. Keep those details with the result so another person can repeat the test and understand its limits.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Production-Grade Local LLM Inference on Apple Silicon (2025)","publisher":"Varun Rajesh et al.","url":"https://arxiv.org/abs/2511.05502"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"GenAI-Perf metrics","publisher":"NVIDIA","url":"https://docs.nvidia.com/deeplearning/triton-inference-server/user-guide/docs/perf_analyzer/genai-perf/README.html"}],"description":"Measure Apple Silicon inference with TTFT, completion latency, throughput, concurrency and task quality. Download a worksheet and define repeatable tests.","audience":"Developers","evidence":{"title":"One local observation is the current performance evidence.","answer":"The product snapshot reports approximately 19–20 output tokens per second on one M4 Max with 36 GB RAM and Qwen3.8 27B at 4-bit, tested September 10. It does not establish fleet performance.","facts":[{"value":"M4 Max · 36 GB","label":"Reported local hardware"},{"value":"4-bit · 27B","label":"Reported model configuration"},{"value":"1 local test","label":"Evidence scope"}],"visual":"benchmark","caption":"Reported by the product team, not independently measured for this guide. The snapshot omits the prompt set, trial count, timing logs and token-counting method.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Provider documentation","url":"https://agents.proxies.sx/compute/"}],"next":{"label":"Download the empty benchmark worksheet","path":"/compute/benchmark-template.csv"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/vs-gpu-cloud","title":"Mac inference compared with GPU cloud and bare-metal Macs","summary":"A managed model endpoint, a remote Mac and a GPU server give you different kinds of access. Decide who needs to install software and what the application must run before you compare the bill.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Rental tiers API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/tiers"},{"title":"Marketplace inventory API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/marketplace"},{"title":"Amazon EC2 Mac instances","publisher":"Amazon Web Services","url":"https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-mac-instances.html"},{"title":"Bare-metal Mac pricing and access","publisher":"MacStadium","url":"https://macstadium.com/pricing"},{"title":"Hosting overview","publisher":"Vast.ai","url":"https://docs.vast.ai/host/hosting-overview"}],"description":"Compare managed Mac inference, bare-metal Macs and GPU cloud by API access, CUDA needs, billing term, model control and operating responsibilities.","audience":"Renters","evidence":{"title":"Choose access and workload before comparing prices.","answer":"A managed Mac endpoint supplies catalog text inference. A bare-metal Mac provides operating-system access under its provider terms. A GPU host must match your runtime, memory and software requirements.","facts":[{"value":"Model API","label":"PROXIES.SX access"},{"value":"No SSH","label":"On a managed rental"},{"value":"30 days","label":"PROXIES.SX term"}],"visual":"comparison","caption":"A workload comparison, not a speed or price ranking. Features of a cloud category vary by provider; verify the individual offer.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"AWS EC2 Mac access and allocation","url":"https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-mac-instances.html"},{"label":"MacStadium access and pricing","url":"https://macstadium.com/pricing"}],"next":{"label":"Compare total costs for equivalent work","path":"/compute/pricing"}},"publishedAt":"2026-09-10","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/provider-troubleshooting","title":"Troubleshoot the MLX compute agent on your Mac","summary":"A loaded background service can still have no model ready to serve. Check the agent, Python environment and local model endpoint separately to find where setup stops.","references":[{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Compute agent source","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/compute-agent.js"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"},{"title":"Download files from the Hub","publisher":"Hugging Face","url":"https://huggingface.co/docs/huggingface_hub/guides/download"},{"title":"Python virtual environments","publisher":"Python Software Foundation","url":"https://docs.python.org/3/library/venv.html"},{"title":"Node.js release schedule","publisher":"Node.js project","url":"https://nodejs.org/en/about/previous-releases"}],"description":"Troubleshoot the current Mac compute installer, pinned Python runtime, model downloads, launchd, approval and offline status with read-only checks.","audience":"Providers","evidence":{"title":"Find the first failed stage before reinstalling.","answer":"An online agent, a serving model, a passed benchmark and an approved listing are separate states. Check runtime logs and the farmer dashboard together before changing a working installation.","facts":[{"value":"60 seconds","label":"Documented heartbeat interval"},{"value":"3 minutes","label":"Missing-heartbeat offline threshold"},{"value":"0.3.1","label":"Reviewed agent release"}],"visual":"steps","steps":[{"title":"Agent cannot start","detail":"Check native arm64 Node 18+, Python 3.10+, macOS 14+ and runtime logs."},{"title":"Model cannot become ready","detail":"Check disk space, download progress, the pinned revision and startup generation errors."},{"title":"Node is paused or offline","detail":"Check power, thermal state, sleep and heartbeat. launchd does not keep a closed laptop awake."},{"title":"Ready but not listed","detail":"Review benchmark status, the countersigned agreement and separate admin approval."}],"caption":"A diagnostic path based on the current agent reference, not a report of faults in your node. Redact keys from logs before requesting support.","sources":[{"label":"Provider documentation","url":"https://agents.proxies.sx/compute/"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Recheck the complete onboarding sequence","path":"/compute/providers"}},"publishedAt":"2026-09-11","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/inference-privacy","title":"Check data privacy before renting an LLM inference endpoint","summary":"A dedicated machine reserves capacity for a customer. To assess privacy, also establish where prompts go, who operates the hardware and what happens to request data after generation.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute operation and pricing reference","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/skill.md"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Compute agent source","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/compute-agent.js"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"},{"title":"LLM01:2025 Prompt Injection","publisher":"OWASP Gen AI Security Project","url":"https://genai.owasp.org/llmrisk/llm01-prompt-injection/"}],"description":"Review prompt retention, supplier access and request handling for dedicated Mac inference. Check the published relay policy, application logs and RAG data.","audience":"Renters","evidence":{"title":"Trace the prompt beyond your application.","answer":"Prompts pass through the platform relay and a supplier-administered Mac. Dedicated tenancy limits who rents that capacity; it does not make the supplier host confidential computing.","facts":[{"value":"~1 hour","label":"Normal completed-content relay scrub"},{"value":"~24 hours","label":"Job-record expiry, with cleanup lag"},{"value":"Supplier-owned","label":"Machine administration"}],"visual":"steps","steps":[{"title":"Your application","detail":"Select permitted text. Check your own logs, stored conversations and fallback destinations."},{"title":"Platform relay","detail":"Temporary relay storage supports request delivery. Published cleanup windows are not zero retention."},{"title":"Supplier Mac","detail":"The supplier administers the hardware. The agent does not intentionally log prompts."},{"title":"Response and saved copies","detail":"The answer returns to your application. Its retention policy is separate from relay cleanup."}],"caption":"Documented data path. Backend deletion and supplier-host behavior were not independently audited. No confidential-computing or compliance certification is established here.","sources":[{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"},{"label":"Compute operation reference","url":"https://agents.proxies.sx/compute/skill.md"}],"next":{"label":"Protect retrieved documents before inference","path":"/compute/rag-inference"}},"publishedAt":"2026-09-11","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/quantization","title":"Choose between 4-bit, 8-bit and bf16 LLM inference","summary":"Fewer bits can reduce model weight storage, but the useful choice depends on the answers your application needs. Compare quantizations with the same tasks before paying for a larger machine.","references":[{"title":"Profiling Large Language Model Inference on Apple Silicon: A Quantization Perspective (2025)","publisher":"Afsara Benazir and Felix Xiaozhu Lin","url":"https://arxiv.org/abs/2508.08531"},{"title":"Model catalog API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"title":"MLX LM runtime documentation","publisher":"MLX project","url":"https://github.com/ml-explore/mlx-lm"},{"title":"The case for 4-bit precision: k-bit Inference Scaling Laws (2023)","publisher":"Tim Dettmers and Luke Zettlemoyer, ICML / PMLR","url":"https://proceedings.mlr.press/v202/dettmers23a.html"}],"description":"Compare 4-bit, 8-bit and bf16 LLM inference on Apple Silicon. Evaluate model quality, weight memory, context, runtime settings and cost per accepted result.","audience":"Renters","evidence":{"title":"Compare capacity without assuming a quality ranking.","answer":"BF16 is a 16-bit format. The catalog pairs its three weight formats with specific memory and context limits; those configurations do not prove a universal speed or quality hierarchy.","facts":[{"value":"4 / 8 / 16 bits","label":"Reviewed weight formats"},{"value":"Same prompts","label":"For meaningful quality comparisons"},{"value":"Separate state","label":"KV cache and working memory"}],"visual":"catalog","caption":"Published catalog requirements, not benchmark results. The 4-bit context limit belongs to this configuration; it is not a general law of quantization.","sources":[{"label":"Live model catalog","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"label":"MLX LM quantized inference","url":"https://github.com/ml-explore/mlx-lm"}],"next":{"label":"Check full model IDs and pinned revisions","path":"/compute/models"}},"publishedAt":"2026-09-11","reviewedAt":"2026-09-11"},{"url":"https://www.proxies.sx/compute/rag-inference","title":"Plan RAG inference hosting on a dedicated Mac endpoint","summary":"A RAG application retrieves relevant source material and supplies it to a language model. A managed chat endpoint can provide the generation step while your application retains the document index and retrieval logic.","references":[{"title":"Dated compute product facts","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/facts.json"},{"title":"Compute provider documentation","publisher":"PROXIES.SX","url":"https://agents.proxies.sx/compute/"},{"title":"Customer documentation in the compute application","publisher":"PROXIES.SX","url":"https://compute.proxies.sx/docs"},{"title":"Model catalog API","publisher":"PROXIES.SX","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"title":"Lost in the Middle: How Language Models Use Long Contexts (2023)","publisher":"Nelson F. Liu et al.","url":"https://arxiv.org/abs/2307.03172"},{"title":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (2020)","publisher":"Patrick Lewis et al.","url":"https://arxiv.org/abs/2005.11401"},{"title":"LLM01:2025 Prompt Injection","publisher":"OWASP Gen AI Security Project","url":"https://genai.owasp.org/llmrisk/llm01-prompt-injection/"}],"description":"Plan RAG with dedicated Apple Silicon inference. Separate retrieval and generation, budget context, test source support and enforce document permissions.","audience":"Developers","evidence":{"title":"Reserve context for the answer as well as the evidence.","answer":"Your application handles ingestion, retrieval and document permissions. The rented endpoint generates text from the excerpts you send. Budget instructions, history, retrieved text and output together.","facts":[{"value":"16,384 / 32,768","label":"Catalog context limits"},{"value":"4,096 tokens","label":"Maximum requested output"},{"value":"Application-owned","label":"Retrieval and access checks"}],"visual":"context","caption":"An editable planning example, not a tokenizer or an acceptance guarantee. The API also enforces a conservative input-byte budget. Retrieved text remains untrusted input.","sources":[{"label":"Live model catalog","url":"https://api.proxies.sx/v1/peer/compute/catalog"},{"label":"Dated product facts","url":"https://agents.proxies.sx/compute/facts.json"}],"next":{"label":"Check API fields, keys and request limits","path":"/compute/api"}},"publishedAt":"2026-09-11","reviewedAt":"2026-09-11"}]}