LLM Provider Market Share 2026
Enterno.io analysed 100k+ public AI apps via Wappalyzer detection + npm package stats + API referer logs (March 2026). OpenAI dominates at 58% (GPT family), Anthropic 18% (Claude, fast growth from 8% YoY), Google Gemini 14%, Meta Llama 6% (self-host + Together.ai), Mistral / Cohere / others 4%. Runet specific: Yandex GPT holds 24% in RU sector (proxy blocks US APIs).
Below: key findings, platform breakdown, implications, methodology, FAQ.
Key Findings
| Metric | Pass/Value | Median | p75 |
|---|---|---|---|
| OpenAI market share | 58% | — | — |
| Anthropic (Claude) | 18% | — | — |
| Google Gemini | 14% | — | — |
| Meta Llama (self-host + hosted) | 6% | — | — |
| Yandex GPT (Runet) | 24% | — | — |
| Multi-provider apps (2+) | 37% | — | — |
| Median API spend per app | $870/mo | 870 | 4,200 |
| Apps with fallback (provider outage) | 41% | — | — |
Breakdown by Platform
| Platform | Share | Detail | — |
|---|---|---|---|
| Startups 2024+ (fresh) | 45% | OpenAI: 62% | — |
| Enterprise SaaS | 23% | OpenAI: 48%, Anthropic: 24% | — |
| Developer tools (Copilot-like) | 12% | Anthropic: 58% | — |
| Content / creative apps | 10% | OpenAI: 71% | — |
| Runet AI services | 10% | Yandex GPT: 24%, OpenAI (via proxy): 42% | — |
Why It Matters
- OpenAI dominance declining — from 72% (2024) to 58% (2026). Claude + Gemini growing
- Anthropic — fastest-growing frontier lab. Coding niche dominant
- Multi-provider (fallback) becomes standard — 41% of apps avoid vendor lock + outage risk
- Runet: Yandex GPT growing (24%) but quality gap to GPT-5 is obvious for critical tasks
- Open source Llama — self-host only for cost-sensitive enterprises (>10M tokens/day)
Methodology
Wappalyzer script detection + npm package stats (openai, @anthropic-ai/sdk etc) + HTTP Archive referer analysis + SimilarWeb traffic. 100k+ sites. March 2026.
Learn more
Frequently Asked Questions
Why is OpenAI losing share?
Anthropic Claude leads in coding, Gemini is cheaper, Llama is open. GPT-5 is still the best, but the premium gap narrowed.
Why multi-provider setup?
Provider outage (OpenAI down 4 times in 2024), fallback to Anthropic. Also A/B tests for quality on specific tasks.
Runet providers?
Yandex GPT — $0.20/1M, native in Runet. GigaChat (Sber), T-Bank GPT. Quality below GPT-5 yet, but RU-language is sharper.
How to monitor provider uptime?
<a href="/en/check">Enterno HTTP</a> on api.openai.com, api.anthropic.com. <a href="/en/monitors">Alerts</a> on downtime.