Open Source AI Popularity Leaderboard
This list ranks AI providers based on the number of unique organizations Scarf sees downloading all open source packages tied to them, across billions of OSS downloads daily.
Need historical data, deeper insights, a live feed, or API access? Contact help@scarf.sh.
Snapshot from August 8, 2026
Foundation model provider packages
Open source SDKs and packages for interacting with providers that develop their own foundation models. Providers are ranked only against other foundation model providers.
| Rank | Provider | Daily trend | Change |
|---|---|---|---|
| 1 | OpenAI | 0 | |
| 2 | Google Gemini | 0 | |
| 3 | Anthropic | 0 | |
| 4 | DeepSeek | 2 | |
| 5 | Alibaba Cloud | 1 | |
| 6 | Mistral AI | 1 | |
| 7 | Zhipu AI (GLM) | 1 | |
| 8 | ByteDance (Doubao) | 1 | |
| 9 | IBM watsonx | 0 | |
| 10 | Voyage AI | 0 | |
| 11 | Cohere | 0 | |
| 12 | Jina AI | 0 | |
| 13 | MiniMax | 1 | |
| 14 | Stability AI | 3 | |
| 15 | Baidu (ERNIE) | 2 | |
| 16 | ElevenLabs | 4 | |
| 17 | Nomic | 2 | |
| 18 | Meta | 0 | |
| 19 | Tencent Hunyuan | 0 | |
| 20 | Kimi | 1 | |
| 21 | Deepgram | 5 |
Inference platform packages
Open source SDKs and packages for interacting with APIs, runtimes, and cloud services used to run or call models. This includes cloud platforms and foundation model providers with hosted APIs, so a provider can appear in both lists and rank differently against the broader peer set.
| Rank | Provider | Daily trend | Change |
|---|---|---|---|
| 1 | OpenAI | 0 | |
| 2 | Amazon Bedrock | 1 | |
| 3 | Google Gemini | 1 | |
| 4 | Anthropic | 1 | |
| 5 | Amazon SageMaker | 3 | |
| 6 | Ollama | 0 | |
| 7 | Microsoft Azure | 0 | |
| 8 | DeepSeek | 2 | |
| 9 | Alibaba Cloud | 1 | |
| 10 | Mistral AI | 1 | |
| 11 | Zhipu AI (GLM) | 2 | |
| 12 | Hugging Face | 1 | |
| 13 | ByteDance (Doubao) | 1 | |
| 14 | IBM watsonx | 1 | |
| 15 | Oracle OCI Generative AI | 2 | |
| 16 | Voyage AI | 0 | |
| 17 | LocalAI | 2 | |
| 18 | Cohere | 0 | |
| 19 | Jina AI | 1 | |
| 20 | GitHub Models | 6 | |
| 21 | MiniMax | 2 | |
| 22 | Stability AI | 6 | |
| 23 | Baidu (ERNIE) | 2 | |
| 24 | OVHcloud AI Endpoints | 2 | |
| 25 | Cloudflare | 0 | |
| 26 | ElevenLabs | 6 | |
| 27 | Nomic | 3 | |
| 28 | Meta | 2 | |
| 29 | Tencent Hunyuan | 2 | |
| 30 | Xinference | 1 | |
| 31 | Kimi | 4 | |
| 32 | LM Studio | 1 | |
| 33 | Deepgram | 7 | |
| 34 | NVIDIA TensorRT / Triton | 0 | |
| 35 | llama.cpp | 8 | |
| 36 | RunPod | 0 | |
| 37 | vLLM | — |
How the leaderboard works
Scarf uses aggregated, organization-level signals from downloads of open source packages to measure how broadly AI providers and platforms are being adopted.
- Open source package signals
- We include downloads of packages that can be confidently associated with a provider or platform and exclude ambiguous activity. The index does not observe API calls or model execution.
- Organization-level ranking
- Providers are ranked by the breadth of observed adoption, not raw downloads, API traffic, model quality, or revenue.
- Daily snapshots
- Rankings update daily. Sparklines show daily unique organizations, while change compares rank with the same category 30 days earlier.
When a package identifies a gateway, router, or model-serving platform, we attribute adoption to that platform. The package usually does not reveal which underlying model was used, so we do not infer one.
This leaderboard is a directional view of observable open source package adoption, not a complete market-share estimate. Package coverage varies by provider and ecosystem. Individual organizations and their activity are not published.
Sign up for our newsletter
Real news and infrequent announcements about what we’re up to at Scarf: product releases, tool updates, open betas, job opportunities, and more.
We care about the protection of your data. Read our Privacy policy.