# The 25 best AI model providers (2026)

> Source: https://agentsindex.ai/categories/ai-model-providers

Updated 15 November 2026 18 prices checked 19 October–14 November 2026 Reviewed by Team AgentsIndex

Sort

[How we check](https://agentsindex.ai/methodology)

25 tools

- [Fireworks AI](https://agentsindex.ai/fireworks-ai) · Inference, model routing and fine-tuning for open models on one OpenAI-compatible endpoint, serverless or in your cloud. · From $0.05 per 1m input tokens Checked 6 Nov · HIPAA SOC 2 · [Visit](https://agentsindex.ai/go/fireworks-ai)
  - MCP server: partly
  - API: yes
  - Open source: no
  - Self-host: partly
- [Gemma](https://agentsindex.ai/gemma) · Open models for generation, retrieval, and safety checks. · Free Checked 28 Oct · Local AI · [Visit](https://agentsindex.ai/go/gemma)
  - MCP server: not listed
  - API: yes
  - Open source: yes
  - Self-host: yes
- [MiniMax](https://agentsindex.ai/minimax) · Frontier lab selling open-weight M3 text, video, speech and music models through token-metered APIs. · From $0.15 per 1m input tokens Checked 29 Oct · [Visit](https://agentsindex.ai/go/minimax)
  - MCP server: yes
  - API: yes
  - Open source: yes
  - Self-host: yes
- [Deepgram](https://agentsindex.ai/deepgram) · A single voice platform covering transcription, speech generation and a voice agent that ties them together. · From $0.0043 per minute Checked 28 Oct · HIPAA · [Visit](https://agentsindex.ai/go/deepgram)
  - MCP server: partly
  - API: yes
  - Open source: partly
  - Self-host: yes
- [nRouter](https://agentsindex.ai/nrouter) · The team behind nRouter runs this listing · LLM gateway billing every model at the provider's list price, with guardrails and budgets checked before spend. · From $5 per minimum credit top up Checked 11 Nov · HIPAA No credit card · [Visit](https://agentsindex.ai/go/nrouter)
  - MCP server: partly
  - API: yes
  - Open source: yes
  - Self-host: not listed
- [OpenAI](https://agentsindex.ai/openai) · OpenAI's GPT models for text, reasoning, realtime voice, images and transcription, with built-in web search tools. · From $0.10 per 1m input tokens Checked 3 Nov · HIPAA No training on your data · [Visit](https://agentsindex.ai/go/openai)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [Vynaris](https://agentsindex.ai/vynaris) · The team behind Vynaris runs this listing · OpenAI-compatible routing that lowers LLM spend with receipts and a model directory. · From $20 a month Checked 25 Oct · [Visit](https://agentsindex.ai/go/vynaris)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: yes
- [01.AI](https://agentsindex.ai/01-ai-yi) · Kai-Fu Lee's foundation-model company, now selling enterprise and sovereign AI deployment around its own models. · On quote Checked 28 Oct · [Visit](https://agentsindex.ai/go/01-ai-yi)
  - MCP server: not listed
  - API: not listed
  - Open source: yes
  - Self-host: not listed
- [Groq](https://agentsindex.ai/groq) · Hosted OpenAI-compatible inference for open-weight models, priced per token. · Free, then $0.075 per 1m input tokens Checked 19 Oct · BYOK SOC 2 · [Visit](https://agentsindex.ai/go/groq)
  - MCP server: no
  - API: yes
  - Open source: no
  - Self-host: no
- [Hume AI](https://agentsindex.ai/hume-ai) · Audio-native decision model that answers plain-language questions about speech and sound with rater-style probabilities. · From $0.47 per 1m input tokens Checked 8 Nov · HIPAA · [Visit](https://agentsindex.ai/go/hume-ai)
  - MCP server: partly
  - API: yes
  - Open source: yes
  - Self-host: not listed
- [Qwen](https://agentsindex.ai/qwen) · Alibaba Cloud's Qwen model family, released as Apache 2.0 open weights and as a hosted Qwen API. · Free, then $0.50 per 1m output tokens Checked 28 Oct · [Visit](https://agentsindex.ai/go/qwen)
  - MCP server: not listed
  - API: not listed
  - Open source: yes
  - Self-host: not listed
- [Hugging Face](https://agentsindex.ai/hugging-face) · A shared hub for models, datasets, and AI apps. · Free, then $9 a month Checked 1 Nov · Freemium · [Visit](https://agentsindex.ai/go/hugging-face)
  - MCP server: yes
  - API: not listed
  - Open source: yes
  - Self-host: not listed
- [Replicate](https://agentsindex.ai/replicate) · AI model inference and fine-tuning through a simple API. · From $0.00 per second Checked 14 Nov · [Visit](https://agentsindex.ai/go/replicate)
  - MCP server: yes
  - API: yes
  - Open source: yes
  - Self-host: not listed
- [DeepSeek](https://agentsindex.ai/deepseek) · AI model platform for chat, mobile, and API workflows. · From $0.003 per 1m input tokens Checked 29 Oct · [Visit](https://agentsindex.ai/go/deepseek)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [Heabsy AI Platform](https://agentsindex.ai/heabsy) · The team behind Heabsy AI Platform runs this listing · OpenAI- and Anthropic-compatible inference API from an EU company, with own models on EEA GPUs and no prompt logging. · From $0.10 per 1m input tokens Checked 2 Nov · No training on your data Zero data retention · [Visit](https://agentsindex.ai/go/heabsy)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [Mistral AI](https://agentsindex.ai/mistral-ai) · Enterprise AI platform for building, deploying, and operating tailored systems. · No list price Checked 4 Nov · EU-hosted · [Visit](https://agentsindex.ai/go/mistral-ai)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [Together AI](https://agentsindex.ai/together-ai) · AI cloud for inference, fine-tuning, and GPU clusters. · No list price Checked 12 Oct · [Visit](https://agentsindex.ai/go/together-ai)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [热核算力 (codexhot)](https://agentsindex.ai/codexhot) · The team behind 热核算力 (codexhot) runs this listing · One API key for routing requests across connected AI models. · From $0.10 per 1m input tokens Checked 3 Nov · [Visit](https://agentsindex.ai/go/codexhot)
  - MCP server: not listed
  - API: yes
  - Open source: not listed
  - Self-host: not listed
- [OpenRouter](https://agentsindex.ai/openrouter) · One API that routes requests across many model providers at list price, with automatic failover. · Free Checked 14 Nov · BYOK Freemium · [Visit](https://agentsindex.ai/go/openrouter)
  - MCP server: yes
  - API: yes
  - Open source: no
  - Self-host: not listed
- [Cohere](https://agentsindex.ai/cohere) · Enterprise model family with private and on-prem deployment, sold alongside its own agent platform. · Free, then $3 per instance hour Checked 1 Nov · Private AI · [Visit](https://agentsindex.ai/go/cohere)
  - MCP server: partly
  - API: yes
  - Open source: partly
  - Self-host: not listed
- [RunPod](https://agentsindex.ai/runpod) · GPU cloud for AI teams running training, inference, and agents. · No list price Checked 15 Aug · [Visit](https://agentsindex.ai/go/runpod)
  - MCP server: yes
  - API: not listed
  - Open source: not listed
  - Self-host: not listed
- [AI21 Labs](https://agentsindex.ai/ai21-labs) · Post-trains specialized models on company data for enterprises, plus usage-priced Jamba foundation model APIs. · From $0.20 per 1m input tokens Checked 4 Nov · [Visit](https://agentsindex.ai/go/ai21-labs)
  - MCP server: not listed
  - API: yes
  - Open source: yes
  - Self-host: yes
- [Poe](https://agentsindex.ai/poe) · Compare AI models, generate media, and search the web in one app. · No list price Checked 12 Oct · [Visit](https://agentsindex.ai/go/poe)
  - MCP server: not listed
  - API: not listed
  - Open source: not listed
  - Self-host: not listed
- [Anthropic](https://agentsindex.ai/anthropic) · AI model provider for Claude and managed-agent infrastructure. · From $0.10 per 1m input tokens Checked 4 Nov · [Visit](https://agentsindex.ai/go/anthropic)
  - MCP server: yes
  - API: not listed
  - Open source: yes
  - Self-host: not listed
- [Microsoft Phi](https://agentsindex.ai/microsoft-phi) · Model access, routing, and governance inside Microsoft Foundry. · No list price Checked 11 Sep · BYOK · [Visit](https://agentsindex.ai/go/microsoft-phi)
  - MCP server: not listed
  - API: not listed
  - Open source: not listed
  - Self-host: not listed
- All 25 side by side · 25 tools in AI model providers · [Compare them](https://agentsindex.ai/categories/ai-model-providers#compared)
  - 2 with a monthly price
  - 4 yes-or-no checks each
  - Prices as of 14 November 2026

The AI models and APIs that apps and agents run on.

$9–$20 a month

First paid plans of the 2 of 25 tools that publish a monthly price.

Vendor pricing pages as of 14 November 2026 n = 2 tools with a monthly price

## What the 25 offer

Show the table

| Offers | Yes | Partly | No or not stated | Of all |
| --- | --- | --- | --- | --- |
| Free to start | 6 | 0 | 19 | 6 of 25 |
| MCP server | 6 | 5 | 14 | 6 of 25 |
| API | 18 | 0 | 7 | 18 of 25 |
| Open source | 10 | 2 | 13 | 10 of 25 |
| Self-host | 5 | 1 | 19 | 5 of 25 |

Each tool’s own site as of 14 November 2026 n = 25 tools

## November in AI model providers

[See where each price was read](#sources)

- Prices checked: **18** of 25 tools, on their own pricing pages
- Last checked: **14 Nov** oldest check 15 Aug

AI model providers compared

| Tool | Starting price | MCP server | API | Open source | Self-host |
| --- | --- | --- | --- | --- | --- |
| [Fireworks AI](https://agentsindex.ai/fireworks-ai) | From $0.05 per 1m input tokens | Partly | Yes | No | Partly |
| [Gemma](https://agentsindex.ai/gemma) | Free plan | Not stated | Yes | Yes | Yes |
| [MiniMax](https://agentsindex.ai/minimax) | From $0.15 per 1m input tokens | Yes | Yes | Yes | Yes |
| [Deepgram](https://agentsindex.ai/deepgram) | From $0.0043 per minute | Partly | Yes | Partly | Yes |
| [nRouter](https://agentsindex.ai/nrouter) | From $5 per minimum credit top up | Partly | Yes | Yes | Not stated |
| [OpenAI](https://agentsindex.ai/openai) | From $0.10 per 1m input tokens | Not stated | Yes | Not stated | Not stated |
| [Vynaris](https://agentsindex.ai/vynaris) | From $20 a month | Not stated | Yes | Not stated | Yes |
| [01.AI](https://agentsindex.ai/01-ai-yi) | Price on request | Not stated | Not stated | Yes | Not stated |
| [Groq](https://agentsindex.ai/groq) | Free plan, then $0.075 per 1m input tokens | No | Yes | No | No |
| [Hume AI](https://agentsindex.ai/hume-ai) | From $0.47 per 1m input tokens | Partly | Yes | Yes | Not stated |
| [Qwen](https://agentsindex.ai/qwen) | Free plan, then $0.50 per 1m output tokens | Not stated | Not stated | Yes | Not stated |
| [Hugging Face](https://agentsindex.ai/hugging-face) | Free plan, then $9 a month | Yes | Not stated | Yes | Not stated |
| [Replicate](https://agentsindex.ai/replicate) | From $0.00 per second | Yes | Yes | Yes | Not stated |
| [DeepSeek](https://agentsindex.ai/deepseek) | From $0.003 per 1m input tokens | Not stated | Yes | Not stated | Not stated |
| [Heabsy AI Platform](https://agentsindex.ai/heabsy) | From $0.10 per 1m input tokens | Not stated | Yes | Not stated | Not stated |
| [Mistral AI](https://agentsindex.ai/mistral-ai) | No list price | Not stated | Yes | Not stated | Not stated |
| [Together AI](https://agentsindex.ai/together-ai) | No list price | Not stated | Yes | Not stated | Not stated |
| [热核算力 (codexhot)](https://agentsindex.ai/codexhot) | From $0.10 per 1m input tokens | Not stated | Yes | Not stated | Not stated |
| [OpenRouter](https://agentsindex.ai/openrouter) | Free plan | Yes | Yes | No | Not stated |
| [Cohere](https://agentsindex.ai/cohere) | Free plan, then $3 per instance hour | Partly | Yes | Partly | Not stated |
| [RunPod](https://agentsindex.ai/runpod) | No list price | Yes | Not stated | Not stated | Not stated |
| [AI21 Labs](https://agentsindex.ai/ai21-labs) | From $0.20 per 1m input tokens | Not stated | Yes | Yes | Yes |
| [Poe](https://agentsindex.ai/poe) | No list price | Not stated | Not stated | Not stated | Not stated |
| [Anthropic](https://agentsindex.ai/anthropic) | From $0.10 per 1m input tokens | Yes | Not stated | Yes | Not stated |
| [Microsoft Phi](https://agentsindex.ai/microsoft-phi) | No list price | Not stated | Not stated | Not stated | Not stated |

**agentsindex.ai** · All 25 AI model providers we list. Each answer is read from the tool's own pages, or tested by us (MCP server, open source); its source is on the tool's page.

[How we check](https://agentsindex.ai/methodology)

This is one of the largest categories in the index. Only 6 of the 25 tools here have a free plan, against 50% across the index. 10 are open source, against 22% across the index.

## Research on AI model providers

- [Guide](https://agentsindex.ai/blog/best-free-ai-agents-real-free-tiers) 1 Jul 2026 11 min Best free AI agents: 10 tools with real free tiers Ten free AI agents with real free tiers and no trial bait, with current pricing and the limits and use cases that matter.

## Price on request only

- 4%: show no price, only “contact sales” (1 of 25 tools), n = 25, 17 November 2026

Counted from the pricing pages of the tools in this category, as our research last read each one. A figure shows only where enough of the tools were read.

## About AI model providers

### What do AI model providers cost?

2 of the 25 publish a monthly price: the first paid plan runs from $9 to $20 a month. The other 23 publish no monthly price. 6 of the 25 have a free plan.

### Which AI model providers are open source?

10 of the 25 listed here are open source: 01.AI (Apache-2.0), AI21 Labs, Anthropic, Gemma, Hugging Face, Hume AI, MiniMax, nRouter and 2 more. 2 more are partly open source: Cohere and Deepgram.

### Which AI model providers have an MCP server?

6 of the 25 have an MCP server: Anthropic, Hugging Face, MiniMax, OpenRouter, Replicate and RunPod. 5 more have one in part: Cohere, Deepgram, Fireworks AI, Hume AI and nRouter.

## Sources and method

Tools rank higher the more we can confirm: a public price, an API and docs, sources, stated limits. [How we check](https://agentsindex.ai/methodology)

Prices are taken from each vendor's pricing page.

- [01.AI](https://agentsindex.ai/01-ai-yi): [01.ai/contact.html](https://www.01.ai/contact.html), checked 28 October 2026
- [AI21 Labs](https://agentsindex.ai/ai21-labs): [ai21.com/pricing](https://www.ai21.com/pricing/), checked 4 November 2026
- [Cohere](https://agentsindex.ai/cohere): [cohere.com/pricing](https://cohere.com/pricing), checked 1 November 2026
- [Deepgram](https://agentsindex.ai/deepgram): [deepgram.com/pricing](https://deepgram.com/pricing), checked 28 October 2026
- [DeepSeek](https://agentsindex.ai/deepseek): [api-docs.deepseek.com/quick_start/pricing](https://api-docs.deepseek.com/quick_start/pricing), checked 29 October 2026
- [Fireworks AI](https://agentsindex.ai/fireworks-ai): [fireworks.ai/pricing](https://fireworks.ai/pricing), checked 6 November 2026

All 18 sources

- [Gemma](https://agentsindex.ai/gemma): [ai.google.dev/gemini-api/docs/pricing](https://ai.google.dev/gemini-api/docs/pricing), checked 28 October 2026
- [Groq](https://agentsindex.ai/groq): [console.groq.com/docs/models](https://console.groq.com/docs/models), checked 19 October 2026
- [Heabsy AI Platform](https://agentsindex.ai/heabsy): [heabsy.com/models](https://heabsy.com/models), checked 2 November 2026
- [Hugging Face](https://agentsindex.ai/hugging-face): [huggingface.co/pricing](https://huggingface.co/pricing), checked 1 November 2026
- [Hume AI](https://agentsindex.ai/hume-ai): [dev.hume.ai/audio-decisions/docs/billing](https://dev.hume.ai/audio-decisions/docs/billing), checked 8 November 2026
- [MiniMax](https://agentsindex.ai/minimax): [platform.minimax.io/docs/pricing/overview.md](https://platform.minimax.io/docs/pricing/overview.md), checked 29 October 2026
- [nRouter](https://agentsindex.ai/nrouter): [nrouter.ai/pricing](https://nrouter.ai/pricing), checked 11 November 2026
- [OpenAI](https://agentsindex.ai/openai): [developers.openai.com/api/docs/pricing](https://developers.openai.com/api/docs/pricing), checked 3 November 2026
- [Qwen](https://agentsindex.ai/qwen): [qwenlm.github.io/blog/qwen-mt](https://qwenlm.github.io/blog/qwen-mt/), checked 28 October 2026
- [Replicate](https://agentsindex.ai/replicate): [replicate.com/pricing](https://replicate.com/pricing), checked 14 November 2026
- [Vynaris](https://agentsindex.ai/vynaris): [vynaris.com/pricing](https://vynaris.com/pricing), checked 25 October 2026
- [热核算力 (codexhot)](https://agentsindex.ai/codexhot): [codexhot.com/model-plaza](https://www.codexhot.com/model-plaza), checked 3 November 2026
