X axis — API/web monthly active usage index (0–100). Y axis — composite benchmark score. Bubble size — estimated developer mindshare %.
Enterprise production-workload usage share (per Menlo Ventures + analyst estimates). Note: consumer/chat traffic share differs sharply — ChatGPT still leads there.
| # | Model | Company | Type | Tier | Ent. Share | Dev Mindshare | YoY Growth | Notable Strength |
|---|
Tier reflects current capability standing (Frontier / Strong / Niche), used in place of Arena Elo given the early-2026 LMArena rebrand that shifted scores independent of model quality. Share and mindshare are analyst estimates.
Open-weight includes Llama, Mistral, Qwen, DeepSeek. Closed includes Anthropic, OpenAI, Google, xAI APIs. Open-weight quality gap has narrowed, but closed still dominates enterprise production.