人工智能

1. 1nomic-ai/gpt4all77,354 stars
2. 2ray-project/ray44,087 stars
3. 3gitleaks/gitleaks30,147 stars
4. 4liguodongiot/llm-action25,186 stars
5. 5Lightning-AI/litgpt13,719 stars
6. 6bentoml/OpenLLM12,561 stars
7. 7openvinotoolkit/openvino11,029 stars
8. 8Tiiny-AI/PowerInfer9,840 stars
9. 9FareedKhan-dev/kimi-k3-in-c9,124 stars
10. 10bentoml/BentoML8,912 stars

Ranked by GitHub stars: how many developers bookmarked the project. Stars measure attention, not quality, so each row also shows forks, open issues, licence and the last push.

1. 1SpaceXAI: Grok 4.20 Multi-Agent2,000,000 tokens
2. 2SpaceXAI: Grok 4.202,000,000 tokens
3. 3Meta: Llama 4 Scout1,310,720 tokens
4. 4OpenAI: GPT-6.1 Sol Pro1,050,000 tokens
5. 5OpenAI: GPT-6.1 Sol Pro (batch)1,050,000 tokens
6. 6OpenAI: GPT-6.1 Sol1,050,000 tokens
7. 7OpenAI: GPT-6.1 Sol (batch)1,050,000 tokens
8. 8OpenAI: GPT-6 Luna Pro1,050,000 tokens
9. 9OpenAI: GPT-6 Luna Pro (batch)1,050,000 tokens
10. 10OpenAI: GPT-6 Luna1,050,000 tokens

The context window is how many tokens (pieces of words) a model takes in one request: a whole codebase, a contract or a book. Each row also shows the input and output price per million...

1. 1OpenAI: gpt-oss-20b$0.02 per million input tokens
2. 2OpenAI: gpt-oss-20b (batch)$0.02 per million input tokens
3. 3OpenAI: GPT-5 Nano (batch)$0.02 per million input tokens
4. 4Meta: Llama 3.2 1B Instruct$0.03 per million input tokens
5. 5Mistral: Mistral Nemo$0.03 per million input tokens
6. 6OpenAI: gpt-oss-120b (batch)$0.03 per million input tokens
7. 7Qwen: Qwen3.7 Flash$0.03 per million input tokens
8. 8Qwen: Qwen3.8 27B$0.04 per million input tokens
9. 9Google: Gemma 3 4B$0.04 per million input tokens
10. 10Anthropic: Claude Haiku 5.5 (batch)$0.05 per million input tokens

Price per million input tokens for models from OpenAI, Anthropic, Google, xAI, Meta, Mistral, DeepSeek, Qwen, Moonshot and MiniMax. Cheaper is not better for every job: compare the...

1. 1Qwen/Qwen3-0.6B28,577,260 downloads (30 days)
2. 2openai-community/gpt217,890,978 downloads (30 days)
3. 3Qwen/Qwen3-4B11,730,390 downloads (30 days)
4. 4Qwen/Qwen3-8B8,809,116 downloads (30 days)
5. 5Qwen/Qwen3-14B8,720,109 downloads (30 days)
6. 6Qwen/Qwen3-Embedding-0.6B8,568,001 downloads (30 days)
7. 7Qwen/Qwen2.5-7B-Instruct6,895,413 downloads (30 days)
8. 8openai/gpt-oss-20b5,933,604 downloads (30 days)
9. 9facebook/opt-125m5,887,680 downloads (30 days)
10. 10Qwen/Qwen2.5-0.5B-Instruct5,751,568 downloads (30 days)

Downloads over the last 30 days on Hugging Face for text-generation models: a sign of what is really used in products and research, small models included.