OpenAI
Maker of the GPT and o-series models, served from OpenAI's own API.
102 models
Explore the model providers available through Nautili.
Connect to supported providers through one Nautili control layer.
Access method depends on provider and account setup.
Maker of the GPT and o-series models, served from OpenAI's own API.
102 models
AWS's managed service for foundation models from several labs, run inside AWS.
100 models
Inference cloud serving open-weight models at per-token prices.
78 models
Microsoft's cloud, serving OpenAI and other models through Azure AI services.
63 models
Google Cloud's AI platform, serving Gemini and partner models with enterprise controls.
54 models
Inference cloud for AI-native startups: serverless open and frontier models behind one OpenAI-format API, plus dedicated endpoints.
41 models
Inference cloud serving open-weight models.
39 models
Privacy-first AI platform serving open-source and third-party models, with anonymised requests and zero data retention on self-hosted models.
37 models
Alibaba Cloud, serving its Qwen models and others.
36 models
Inference cloud serving open-weight models and GPU instances.
35 models
Maker of the Claude models, served from Anthropic's own API.
31 models
Google's developer API for Gemini models.
30 models
GPU cloud with a serverless inference service for LLM and multimodal models, scaling from zero to dedicated clusters.
27 models
Inference and training cloud for open-weight models.
24 models
GPU cloud with managed inference.
23 models
Paris-based lab serving its open-weight and commercial Mistral models.
23 models
AI cloud with a hosted model inference service.
21 models
One API for 400+ image, video, audio and chat models.
20 models
Workers AI, serverless inference on Cloudflare's global network.
18 models
Singapore-based AI cloud serving models such as DeepSeek and KAT-Coder through its Vanchin platform, alongside video cloud services.
18 models
Cloud provider with serverless inference for open models.
17 models
Baidu AI Cloud, serving the ERNIE models.
15 models
Chinese lab behind the GLM models.
15 models
Fast models for coding agents, including Fast Apply for merging code edits at thousands of tokens per second.
14 models
Platform for training, deploying and improving your own models, with hosted open-weight inference through one OpenAI-compatible API.
14 models
Inference platform for serving open and custom models.
13 models
Open-weight inference provider serving DeepSeek, Kimi and GLM models at high volume.
13 models
Anthropic's Claude models served from Anthropic's platform running on AWS.
12 models
Enterprise AI company behind the Command and Embed models.
12 models
Decentralised private inference network running open models on contributed Apple Silicon Macs, via an OpenAI-compatible API.
12 models
Infrastructure for long-horizon agents: open models like Kimi, GLM and DeepSeek through OpenAI- and Anthropic-compatible APIs, plus Sailboxes VMs.
12 models
Inference platform that learns your workload and continuously tunes dedicated model deployments for speed and cost.
12 models
Fast inference platform for open-weight models.
11 models
Maker of the Grok models.
11 models
Inference service running large language models on high-end hardware, with free and paid tiers.
10 models
Serverless compute platform for running models.
10 models
NVIDIA-backed GPU cloud for AI workloads from Lintasarta Cloudeka, an Indonesian cloud provider.
9 models
Inference infrastructure for AI-native teams: fast serving for open-source and tuned models such as GPT-OSS and Kimi.
9 models
Startup whose AI agent writes and tunes GPU kernels, serving open-weight models like Kimi, GLM and Gemma with them.
9 models
Decentralised, per-token inference for open models such as Qwen, GLM, Kimi and Llama, running on the Akash Network.
8 models
Decentralised serverless compute hosting open-source LLMs, image, video and speech models, built on Bittensor.
8 models
AI cloud with managed inference.
8 models
Lab building multimodal models.
8 models
Serverless model APIs and dedicated deployments with a compiler-based optimisation layer for cheaper inference.
7 models
Models available through Nautili.
7 models
Maker of the Llama models, served through Meta's developer API.
7 models
Modular's hosted inference: OpenAI-compatible shared and dedicated endpoints on the MAX serving stack.
7 models
Tencent Cloud, serving the Hunyuan models.
7 models
OpenAI-compatible API with model tiers from fast to frontier, plus autonomous email agents.
6 models
Low-latency inference on Groq's own LPU chips.
6 models
Spanish provider running sovereign inference in Europe: pay-per-token API for open models like Qwen, DeepSeek and Llama.
6 models
AI hardware company selling Blackhole cards and Galaxy servers; inference here runs on Tenstorrent's own chips.
6 models
Inference platform focused on fast, efficient serving.
5 models
Search-grounded Sonar models with live web results.
5 models
Models available through Nautili.
5 models
Decentralised GPU network pooling capacity from independent data centres, with an inference API for open models.
4 models
Chinese lab behind the Kimi models.
4 models
Private AI inference you can verify: open-weight models in hardware-isolated TEEs, with attestation on every response.
4 models
NVIDIA's hosted API catalog for its own and partner models.
4 models
Models available through Nautili.
4 models
Models available through Nautili.
4 models
Tokyo-based lab building frontier AI in Japan, including the Japanese-specialised Namazu model and the Fugu orchestrator.
4 models
Fast inference on SambaNova's own RDU chips.
4 models
Models available through Nautili.
3 models
Models available through Nautili.
3 models
GPU cloud offering dedicated single-tenant NVIDIA B200 and B300 bare-metal servers for training and inference.
3 models
Lab building efficient Liquid Foundation Models.
3 models
Lab behind the Nex-N2.5 model family, built to operate computers, browsers and code from visual feedback.
3 models
Reworks open models to run efficiently on non-GPU accelerators, serving them through OpenRouter.
3 models
ByteDance's Seed models, including Seedance video and Seedream image generation, served through BytePlus ModelArk.
3 models
Chinese lab behind the Step models.
3 models
Benchmarks frontier and open models continuously and serves Pareto, a blended model that returns the strongest answer.
3 models
Xiaomi's MiMo API platform, serving its MiMo models.
3 models
Verifiable AI inference on an open network: open models via OpenAI- and Anthropic-compatible APIs, with Proof of Logits verification.
2 models
High-speed inference on Cerebras wafer-scale chips.
2 models
Real-time AI lab behind the Lucy video-editing and Oasis world models, with its own inference and training stack.
2 models
Chinese lab behind the DeepSeek open-weight models.
2 models
Chinese lab behind the MiniMax text and multimodal models.
2 models
Model layer for physical AI: the Mk1.5 and Isaac vision models, available by API or licensed weights.
2 models
Builds task-specific small language models.
2 models
Korean lab behind the Solar models.
2 models
Models available through Nautili.
1 model
Models available through Nautili.
1 model
Models available through Nautili.
1 model
Models available through Nautili.
1 model
AI research and product lab behind the Inkling open-weight models and the Tinker platform.
1 model
Route across supported providers through one control layer.