The problem
Companies want AI models they own and can improve, without building a training and GPU infrastructure team
What they do
API to fine-tune, RL-train and serve your own open-weight AI models
Traction Early: API live since Jun 2026; named users incl. Perplexity, Abacus AI; est. revenue $3-7M (Seedtable)
Hiring trend
- Open roles
- 19
- Ops roles
- 2
- Headcount
- ~20-23
New: first roles posted May 2026, 19 open now; headcount ~20 in Aug to ~22-23 (LinkedIn)
Hiring in
- Palo Alto
- Austin
- SF Bay Area
Funding
- Last round
- $1.1B
- Stage
- Series A
- Date
- Aug 2026
Led by General Catalyst, AMP PBC
- Hiring+1.519 open roles on the live job board (10–29 roles = +1.5)
- Funding recency+2Series A · $1.1B, Aug 2026 (2 months ago): 6 months or less = +1.5, $100M+ round +0.5
- Traction+0.75Some growth: Early: API live since Jun 2026; named users incl. Perplexity, Abacus AI; est. revenue $3-7M (Seedtable)
- Red flags0None found in the last 24 months
- Total4.251.5 + 2 + 0.75 − 0 = 4.25 → Warm-up
Seed rule (editor's call): early-stage without strong traction starts in Warm-up (Race by score alone).
Distance (Unproven) is a separate read on how long the lead can last. It doesn't change the score.
Distance
Unproven
Four months old with ~20 staff and a just-launched API; competes in a crowded post-training and inference field (Together, Fireworks and others) where price cuts are easy to match, so staying power can't be judged yet despite a deep war chest.
Sources: TechCrunch: General Catalyst leads $1.1B round into 2-month-old River AI (Aug 2026) ↗ · AI Tamer: River Inference lists open models 20% under labs' rates (Oct 2026) ↗
Open ops roles · 2
Last checked Oct 25, 2026