Developer maps the entire inference + fine-tuning provider landscape with tradeoffs
Present-Jelly9941 · reddit · 2026-09-09
A developer who spent weeks shopping for inference + fine-tuning published a full landscape map. Key thesis: grouping matters more than names — vendors within a group are substitutable, across groups they aren't. Groups cover token-API+FT (Fireworks, Together), FT-first (Predibase, OpenPipe), managed deploy (Baseten, Replicate, Modal), raw GPU (RunPod, Lambda, CoreWeave), speed specialists (Groq, Cerebras), routers (OpenRouter), and Akka's cost-per-task routing. Self-hosting floor: vLLM/SGLang + LoRAX, trained with Axolotl/Unsloth/TRL. Per-token vs cost-per-task is a bet on workload narrowness, not a benchmark decision.
More from Venture
- Series B AI CRM firm scrapes job posts to map rivals' CRMs for targeted outreach — MartinGTobias · 2026-09-09
- Series A bar surges: median round up from $4.5M to $19.4M as lead VCs drop to ~50 a year — MartinGTobias · 2026-09-09
- Stanford's Christopher Manning: Silicon Valley has a 'naive belief in gurus', calls AI funding levels 'manifestly crazy' — chrmanning · 2026-09-09
- Three focal points for AI rollup companies: asset selection, talent compensation, and incentive alignment — curious_vii · 2026-09-09
- Ramp data: no-ZDR Fable 5.1 hits 22.5% of enterprise spend, ZDR a hard requirement — zephyr_z9 · 2026-09-09
- 2026 AI rollup market map grades 45 firms; none yet earns third-party-verified top marks — curious_vii · 2026-09-09