[{"data":1,"prerenderedAt":81},["ShallowReactive",2],{"$f11mfxiaqi9mtf":3},{"site":4,"app":16,"alternatives":57,"compose":54,"log":77,"changelog":78},{"hourly_rate_usd":5,"break_even_window_mo":6,"reference_env":7,"attribution":8},20,18,"Fresh Debian 12, Docker + Compose preinstalled, 2 GB RAM \u002F 2 vCPU ($5-class VPS)",{"caniselfhostit":9},{"name":10,"url":11,"repo":12,"license":13,"author":14,"note":15},"caniselfhostit.com","https:\u002F\u002Fcaniselfhostit.com","https:\u002F\u002Fgithub.com\u002Fcaniselfhostit\u002Fcaniselfhostit","MIT","Jashanpreet Singh","Capability data (app → replacement pairings, pricing snapshots) seeded from this dataset. They answer 'can you run it, and how' — go there for guided installs.",{"id":17,"name":18,"domain":19,"category":20,"price_usd_mo":21,"price_plan":22,"price_unit":23,"price_source":24,"price_checked":25,"links":26,"jobs":28,"verdict":32,"alternatives":33,"provenance":35,"verdict_reason":38,"what_you_lose":39,"economics":41,"verified":44,"derived":52},"midjourney","Midjourney","midjourney.com","ai-tools",10,"Basic","flat","https:\u002F\u002Fdocs.midjourney.com\u002Fhc\u002Fen-us\u002Farticles\u002F27870484040333","2026-08-07",{"caniselfhostit":27},"https:\u002F\u002Fcaniselfhostit.com\u002Fmidjourney",[29,30,31],"text-to-image generation","style control","upscaling & variations","NOT_REALLY",[34],"invokeai",{"capability_seed":36,"decision_layer":37},"caniselfhostit","original","InvokeAI + open models on your own GPU is a genuine creative stack — for people who own the GPU for other reasons. Renting one to avoid a $10\u002Fmo subscription is negative-sum, and Midjourney's model quality is the product anyway. Same GPU-economics pattern as Copilot.",[40],"Midjourney's models and the zero-setup pipeline",{"vps_share_usd_mo":42,"storage_usd_mo":42,"maint_min_mo":43},0,30,{"by":45,"at":46,"protocol":47,"setup_min":48,"assistant":49,"evidence":50},"zernonia","2026-08-10","v1",15,"claude-code",[51],"logs\u002Fmidjourney.md",{"hourly_rate_usd":5,"self_host_cost_usd_mo":21,"net_saving_usd_mo":42,"setup_cost_usd":53,"break_even_months":54,"break_even_date":54,"markup_index":55,"primary_abandoned":56,"verdict_stale":56},5,null,1,false,[58],{"id":34,"name":59,"tagline":60,"repo":61,"site":62,"license":63,"category":20,"resources":64,"provenance":67,"metrics":68},"InvokeAI","Text-to-image on a card you own, with the checkpoint, the gallery and the electricity bill all in one building.","invoke-ai\u002FInvokeAI","https:\u002F\u002Finvoke.ai","Apache-2.0",{"ram_min_mb":65,"disk_gb":66,"arm64":56,"containers":55,"external_db":56},16384,40,{"capability_seed":36,"decision_layer":37},{"id":34,"stars":69,"last_commit":70,"latest_release":71,"docker_pulls":54,"open_issues":75,"fetched_at":76},27829,"2026-08-09T18:28:32Z",{"tag":72,"published_at":73,"url":74},"v6.13.7","2026-07-21T23:48:34Z","https:\u002F\u002Fgithub.com\u002Finvoke-ai\u002FInvokeAI\u002Freleases\u002Ftag\u002Fv6.13.7",394,"2026-08-11T03:55:23.678Z","# Midjourney — verdict log\n\n- **Date:** 2026-08-10\n- **Verdict:** NOT REALLY — keep paying\n- **Protocol:** v1 (2GB reference env, Docker preinstalled, AI assistant: claude-code)\n\n## Why no compose file\n\n- This verdict is decided by economics\u002Fcapability, not setup effort — the rubric doesn't require a timed install to conclude the subscription should stay (or that the replacement's cost lives elsewhere, e.g. GPUs, egress, or an installer script).\n\n## The call\n\nInvokeAI + open models on your own GPU is a genuine creative stack — for people who own the GPU for other reasons. Renting one to avoid a $10\u002Fmo subscription is negative-sum, and Midjourney's model quality is the product anyway. Same GPU-economics pattern as Copilot.\n",[79],{"date":46,"app":17,"from":54,"to":32,"reason":80},"Initial verdict: InvokeAI + open models on your own GPU is a genuine creative stack — for people who own the GPU for other reas.",1786459421448]