# Nexus Gateway

> Own your AI Gateway. Access 100+ AI models in one place. Give your people every major model on the web, desktop, Slack, mobile, CLI and API, with budgets, model rules and audit, and bring your own keys and licenses free.

Nexus is an AI gateway a company owns: every major model (Claude, GPT, Gemini and more; 100+ models as the catalog grows) in the web app, desktop, Slack, mobile, CLI and API, under one set of controls (nested organizational units, budgets, model allowlists, audit). Bring your own API keys (free up to $25,000 a month of list-price usage, then 5%) and your Claude and ChatGPT plans (free). The Auto Router sends each step of a request to the cheapest model that can do it.

## Docs for agents

- [Overview](https://nexus.inc/index.md): the whole site in Markdown
- [Pricing](https://nexus.inc/pricing.md): plans, allowances and how we compare
- [Full text](https://nexus.inc/llms-full.txt): every page in one file

## Product

### Interfaces

- [Web app](https://nexus.inc/product/web): Every model, one chatbox. Open nexus.inc and ask any model your company allows. Nothing to install.
- [Desktop app](https://nexus.inc/product/desktop): Coding agents on your Mac. Claude Code and Codex in your own repos, through the gateway.
- [Mac widget](https://nexus.inc/product/widget): One keystroke from the menu bar. Ask, approve and watch your agents without opening a window.
- [Mobile apps](https://nexus.inc/product/mobile): Approve from anywhere. Shared answers on a tablet. Approvals and alerts on a phone.
- [Slack agent](https://nexus.inc/product/slack): Every AI, natively in your Slack. Tag @Nexus in any channel. Auto picks the cheapest model for each step, or you name the one you want.
- [CLI and API](https://nexus.inc/product/cli-api): Built for engineers and agents. nexus claude and nexus codex on your keys and budgets. One endpoint for every model, with the cost on every response.

### Features

- [Artifacts](https://nexus.inc/product/artifacts): Share the answer, not a screenshot. Charts, tables and code in one clean page, with the sources and the models that made it. Share it with a team, a person or a link.
- [Wiki](https://nexus.inc/product/wiki): A living wiki of your code. Wiki writes it from your repositories and keeps it current.
- [Review](https://nexus.inc/product/review): A first review on every pull request. Blocking issues first, fixes in one click, routed to the right owner.
- [Cloud Agents](https://nexus.inc/product/cloud-agents): Many agents, working in the cloud. Each agent runs in its own workspace and keeps going when your laptop closes.
- [Synced sessions](https://nexus.inc/product/sync): Start anywhere, pick up anywhere. Start in the terminal, continue in the browser, approve from your phone.

### Platform

- [Enterprise controls](https://nexus.inc/product/controls): Enterprise-grade controls, on your own gateway. Every request from every surface passes one gateway that checks who is asking, the budget, the models and the data.
- [Organization](https://nexus.inc/product/organization): Built for how large companies are organized. Nested units as deep as you need: divisions, teams, people and agents.
- [Budgets and allowlists](https://nexus.inc/product/budgets): Budgets and allowlists, per unit. A budget for every unit and a model policy that allows or blocks each model.
- [Connectors](https://nexus.inc/product/connectors): Connect once, use everywhere. Every model, agent and surface can use the tools you connect.
- [Bring your own keys](https://nexus.inc/product/byok): Your keys and licenses. Free. Bring the API keys and AI plans you already own, with our controls on top.

### Models

- [Explorer](https://nexus.inc/product/explorer): Find the right model, and know it. Every model ranked, priced and explained, so anyone on the team can choose.
- [Auto Router](https://nexus.inc/product/auto-router): The right model for each task. A complexity-aware router: each step goes to the cheapest model that can do it well, inside your policies, to cut your AI bill.
- [Benchmarks](https://nexus.inc/product/benchmarks): Proof before you switch. Auto against every model, on public benchmarks and on your own tasks.

## Explorer

Every AI model, ranked by LMArena votes and Vals AI benchmarks, with list prices, context windows and Hugging Face downloads. 440 models from 42 labs, updated 2026-10-24. Data: [explorer.json](https://nexus.inc/explorer.json).

- [Explorer](https://nexus.inc/models): all models, score against price, leaders
- [Rankings](https://nexus.inc/rankings): [Text](https://nexus.inc/rankings/lmarena-text), [Coding](https://nexus.inc/rankings/lmarena-coding), [Math](https://nexus.inc/rankings/lmarena-math), [Creative writing](https://nexus.inc/rankings/lmarena-creative-writing), [Instruction following](https://nexus.inc/rankings/lmarena-instruction-following), [Hard prompts](https://nexus.inc/rankings/lmarena-hard-prompts), [Multi-turn](https://nexus.inc/rankings/lmarena-multi-turn), [Long queries](https://nexus.inc/rankings/lmarena-longer-query), [Expert](https://nexus.inc/rankings/lmarena-expert), [Legal and government](https://nexus.inc/rankings/lmarena-legal-government), [Medicine and healthcare](https://nexus.inc/rankings/lmarena-medicine-healthcare), [Business and finance](https://nexus.inc/rankings/lmarena-business-management-financial-operations), [Chinese](https://nexus.inc/rankings/lmarena-chinese), [Spanish](https://nexus.inc/rankings/lmarena-spanish), [German](https://nexus.inc/rankings/lmarena-german), [French](https://nexus.inc/rankings/lmarena-french), [Japanese](https://nexus.inc/rankings/lmarena-japanese), [WebDev](https://nexus.inc/rankings/lmarena-webdev), [Vision](https://nexus.inc/rankings/lmarena-vision), [Hugging Face downloads](https://nexus.inc/rankings/hugging-face)
- [Benchmarks](https://nexus.inc/benchmarks): [Vals Index](https://nexus.inc/benchmarks/vals-index), [Terminal-Bench 4.0](https://nexus.inc/benchmarks/terminal-bench-4-0), [SWE-bench](https://nexus.inc/benchmarks/swe-bench), [Vibe Code Bench](https://nexus.inc/benchmarks/vibe-code-bench), [Code Migration](https://nexus.inc/benchmarks/code-migration), [LiveCodeBench](https://nexus.inc/benchmarks/livecodebench), [Finance Agent](https://nexus.inc/benchmarks/finance-agent), [Finance Index](https://nexus.inc/benchmarks/finance-index), [Tax Agent Bench](https://nexus.inc/benchmarks/tax-agent-bench), [Legal Index](https://nexus.inc/benchmarks/legal-index), [Legal Research Bench](https://nexus.inc/benchmarks/legal-research-bench), [MedCode](https://nexus.inc/benchmarks/medcode), [MedQA](https://nexus.inc/benchmarks/medqa), [GPQA Diamond](https://nexus.inc/benchmarks/gpqa-diamond), [MMLU-Pro](https://nexus.inc/benchmarks/mmlu-pro), [AIME](https://nexus.inc/benchmarks/aime), [MMMU](https://nexus.inc/benchmarks/mmmu), [Terminal-Bench Science](https://nexus.inc/benchmarks/terminal-bench-science)
- [Compare models](https://nexus.inc/compare)
- [Data](https://nexus.inc/data): open vs proprietary, the gap between them, labs and countries of the top 20, releases, prices

### Top models

- [Gemini 4 Argon](https://nexus.inc/models/gemini-4-argon): Gemini 4 Argon is a proprietary model from Google, released Oct 13, 2026. It ranks #1 of 351 on the LMArena text leaderboard with a rating of 1526, and #3 of 27 on the Vals Index with 69.24%. It costs $2 per million input tokens and $10 per million output tokens, with a 1M-token context window.
- [Claude Opus 5.5](https://nexus.inc/models/claude-opus-5-5): Claude Opus 5.5 is a proprietary model from Anthropic, released Sep 22, 2026. It ranks #2 of 351 on the LMArena text leaderboard with a rating of 1516, and #2 of 27 on the Vals Index with 70.18%. It costs $4 per million input tokens and $20 per million output tokens, with a 1M-token context window.
- [Claude Fable 5.1](https://nexus.inc/models/claude-fable-5-1): Claude Fable 5.1 is a proprietary model from Anthropic, released Sep 1, 2026. It ranks #3 of 351 on the LMArena text leaderboard with a rating of 1513. It costs $10 per million input tokens and $50 per million output tokens, with a 1M-token context window.
- [Claude Opus 5](https://nexus.inc/models/claude-opus-5): Claude Opus 5 is a proprietary model from Anthropic, released Jul 24, 2026. It ranks #4 of 351 on the LMArena text leaderboard with a rating of 1508. It costs $5 per million input tokens and $25 per million output tokens, with a 1M-token context window.
- [Claude Opus 4.6](https://nexus.inc/models/claude-opus-4-6): Claude Opus 4.6 is a proprietary model from Anthropic, released Feb 4, 2026. It ranks #6 of 351 on the LMArena text leaderboard with a rating of 1502. It costs $5 per million input tokens and $25 per million output tokens, with a 1M-token context window.
- [Muse Wonder](https://nexus.inc/models/muse-wonder): Muse Wonder is a proprietary model from Meta, released Oct 12, 2026. It ranks #7 of 351 on the LMArena text leaderboard with a rating of 1501, and #6 of 27 on the Vals Index with 64.88%. It costs $4 per million input tokens and $20 per million output tokens, with a 1M-token context window.
- [Gemini 3.8 Flash](https://nexus.inc/models/gemini-3-8-flash): Gemini 3.8 Flash is a proprietary model from Google, released Sep 2, 2026. It ranks #8 of 351 on the LMArena text leaderboard with a rating of 1498, and #15 of 27 on the Vals Index with 55.95%. It costs $0.82 per million input tokens and $4.13 per million output tokens, with a 1M-token context window.
- [MiMo-V2.6-Pro](https://nexus.inc/models/mimo-v2-6-pro): MiMo-V2.6-Pro is an open-weight model from Xiaomi, released Sep 21, 2026. It ranks #11 of 351 on the LMArena text leaderboard with a rating of 1491, and #14 of 27 on the Vals Index with 57.81%. It costs $0.43 per million input tokens and $0.87 per million output tokens, with a 1.1M-token context window.
- [Muse Spark 1.3](https://nexus.inc/models/muse-spark-1-3): Muse Spark 1.3 is a proprietary model from Meta, released Sep 2, 2026. It ranks #10 of 351 on the LMArena text leaderboard with a rating of 1491. It costs $1.25 per million input tokens and $4.25 per million output tokens, with a 1M-token context window.
- [Claude Fable 5](https://nexus.inc/models/claude-fable-5): Claude Fable 5 is a proprietary model from Anthropic, released Jun 9, 2026. It ranks #12 of 351 on the LMArena text leaderboard with a rating of 1491. It costs $10 per million input tokens and $50 per million output tokens, with a 1M-token context window.
- [Claude Opus 4.7](https://nexus.inc/models/claude-opus-4-7): Claude Opus 4.7 is a proprietary model from Anthropic, released Apr 16, 2026. It ranks #13 of 351 on the LMArena text leaderboard with a rating of 1490. It costs $5 per million input tokens and $25 per million output tokens, with a 1M-token context window.
- [Gemini 3.7 Flash](https://nexus.inc/models/gemini-3-7-flash): Gemini 3.7 Flash is a proprietary model from Google, released Aug 13, 2026. It ranks #14 of 351 on the LMArena text leaderboard with a rating of 1490. It costs $0.75 per million input tokens and $3.75 per million output tokens, with a 1M-token context window.
- [Qwen3.8-Max](https://nexus.inc/models/qwen3-8-max): Qwen3.8-Max is a proprietary model from Alibaba, released Sep 3, 2026. It ranks #15 of 351 on the LMArena text leaderboard with a rating of 1484, and #11 of 27 on the Vals Index with 59.05%. It costs $2 per million input tokens and $6 per million output tokens, with a 1M-token context window.
- [Gemini 3.5 Flash](https://nexus.inc/models/gemini-3-5-flash): Gemini 3.5 Flash is a proprietary model from Google, released May 19, 2026. It ranks #18 of 351 on the LMArena text leaderboard with a rating of 1482. It costs $1.50 per million input tokens and $9 per million output tokens, with a 1M-token context window.
- [Muse Spark 1.2](https://nexus.inc/models/muse-spark-1-2): Muse Spark 1.2 is a proprietary model from Meta, released Aug 5, 2026. It ranks #17 of 351 on the LMArena text leaderboard with a rating of 1482. It costs $1.25 per million input tokens and $4.25 per million output tokens, with a 1M-token context window.
- [Gemini 3.1 Pro Preview](https://nexus.inc/models/gemini-3-1-pro-preview): Gemini 3.1 Pro Preview is a proprietary model from Google, released Feb 19, 2026. It ranks #19 of 351 on the LMArena text leaderboard with a rating of 1480. It costs $2 per million input tokens and $12 per million output tokens, with a 1M-token context window.
- [Gemini 3.6 Flash](https://nexus.inc/models/gemini-3-6-flash): Gemini 3.6 Flash is a proprietary model from Google, released Jul 21, 2026. It ranks #20 of 351 on the LMArena text leaderboard with a rating of 1480. It costs $0.75 per million input tokens and $3.75 per million output tokens, with a 1M-token context window.
- [Gemini 3 Pro](https://nexus.inc/models/gemini-3-pro): Gemini 3 Pro is a proprietary model from Google. It ranks #21 of 351 on the LMArena text leaderboard with a rating of 1480.
- [Muse Spark 1.1](https://nexus.inc/models/muse-spark-1-1): Muse Spark 1.1 is a proprietary model from Meta, released Jul 16, 2026. It ranks #22 of 351 on the LMArena text leaderboard with a rating of 1478. It costs $1.25 per million input tokens and $4.25 per million output tokens, with a 1M-token context window.
- [Qwen3.7-Max-Preview](https://nexus.inc/models/qwen3-7-max-preview): Qwen3.7-Max-Preview is a proprietary model from Alibaba, released May 21, 2026. It ranks #24 of 351 on the LMArena text leaderboard with a rating of 1475. It costs $1.48 per million input tokens and $4.42 per million output tokens, with a 1M-token context window.
- [Kimi K3](https://nexus.inc/models/kimi-k3): Kimi K3 is an open-weight model from Moonshot AI, released Jul 16, 2026. It ranks #25 of 351 on the LMArena text leaderboard with a rating of 1475, and #20 of 27 on the Vals Index with 52.95%. It costs $0.24 per million input tokens and $11 per million output tokens, with a 1M-token context window.
- [Muse Spark](https://nexus.inc/models/muse-spark): Muse Spark is a proprietary model from Meta. It ranks #26 of 351 on the LMArena text leaderboard with a rating of 1474.
- [GLM-5.3](https://nexus.inc/models/glm-5-3): GLM-5.3 is an open-weight model from Z.ai, released Aug 18, 2026. It ranks #27 of 351 on the LMArena text leaderboard with a rating of 1474, and #10 of 27 on the Vals Index with 59.73%. It costs $0.079 per million input tokens and $1.40 per million output tokens, with a 1M-token context window.
- [Claude Sonnet 5.5](https://nexus.inc/models/claude-sonnet-5-5): Claude Sonnet 5.5 is a proprietary model from Anthropic, released Sep 28, 2026. It ranks #28 of 351 on the LMArena text leaderboard with a rating of 1473, and #1 of 27 on the Vals Index with 70.85%. It costs $2 per million input tokens and $10 per million output tokens, with a 1M-token context window.
- [GLM-5.3-Flash](https://nexus.inc/models/glm-5-3-flash): GLM-5.3-Flash is an open-weight model from Z.ai, released Aug 26, 2026. It ranks #29 of 351 on the LMArena text leaderboard with a rating of 1472, and #22 of 27 on the Vals Index with 50.86%. It costs $0.15 per million input tokens and $0.50 per million output tokens, with a 1M-token context window.
- [GPT-5.5](https://nexus.inc/models/gpt-5-5): GPT-5.5 is a proprietary model from OpenAI, released Apr 24, 2026. It ranks #30 of 351 on the LMArena text leaderboard with a rating of 1472. It costs $5 per million input tokens and $30 per million output tokens, with a 1.1M-token context window.
- [Qwen3.5-Max-Preview](https://nexus.inc/models/qwen3-5-max-preview): Qwen3.5-Max-Preview is a proprietary model from Alibaba. It ranks #31 of 351 on the LMArena text leaderboard with a rating of 1470.
- [GLM-5.2](https://nexus.inc/models/glm-5-2): GLM-5.2 is an open-weight model from Z.ai, released Jun 16, 2026. It ranks #32 of 351 on the LMArena text leaderboard with a rating of 1470. It costs $0.034 per million input tokens and $7 per million output tokens, with a 1M-token context window.
- [GPT-5.4](https://nexus.inc/models/gpt-5-4): GPT-5.4 is a proprietary model from OpenAI, released Mar 5, 2026. It ranks #33 of 351 on the LMArena text leaderboard with a rating of 1469. It costs $2.50 per million input tokens and $15 per million output tokens, with a 1.1M-token context window.
- [ERNIE 5.1](https://nexus.inc/models/ernie-5-1): ERNIE 5.1 is a proprietary model from Baidu. It ranks #34 of 351 on the LMArena text leaderboard with a rating of 1468.
- [Gemini 3 Flash](https://nexus.inc/models/gemini-3-flash): Gemini 3 Flash is a proprietary model from Google, released Dec 17, 2025. It ranks #36 of 351 on the LMArena text leaderboard with a rating of 1466. It costs $0.50 per million input tokens and $3 per million output tokens, with a 1M-token context window.
- [DeepSeek-V4.1-Flash-20260910](https://nexus.inc/models/deepseek-v4-1-flash-20260910): DeepSeek-V4.1-Flash-20260910 is an open-weight model from DeepSeek, released Oct 17, 2026. It ranks #37 of 351 on the LMArena text leaderboard with a rating of 1464, and #9 of 27 on the Vals Index with 61.26%. It costs $0.15 per million input tokens and $0.60 per million output tokens, with a 1M-token context window.
- [MiMo-V2.5-Pro](https://nexus.inc/models/mimo-v2-5-pro): MiMo-V2.5-Pro is an open-weight model from Xiaomi, released Apr 22, 2026. It ranks #38 of 351 on the LMArena text leaderboard with a rating of 1464. It costs $0.30 per million input tokens and $0.61 per million output tokens, with a 1.1M-token context window.
- [GLM-5.1](https://nexus.inc/models/glm-5-1): GLM-5.1 is an open-weight model from Z.ai, released Apr 7, 2026. It ranks #39 of 351 on the LMArena text leaderboard with a rating of 1463. It costs $1.40 per million input tokens and $4.40 per million output tokens, with a 200K-token context window.
- [Claude Opus 4.8](https://nexus.inc/models/claude-opus-4-8): Claude Opus 4.8 is a proprietary model from Anthropic, released May 27, 2026. It ranks #40 of 351 on the LMArena text leaderboard with a rating of 1461. It costs $5 per million input tokens and $25 per million output tokens, with a 1M-token context window.
- [Claude Sonnet 4.6](https://nexus.inc/models/claude-sonnet-4-6): Claude Sonnet 4.6 is a proprietary model from Anthropic, released Feb 17, 2026. It ranks #41 of 351 on the LMArena text leaderboard with a rating of 1458. It costs $3 per million input tokens and $15 per million output tokens, with a 1M-token context window.
- [Gemini 2.5 Pro](https://nexus.inc/models/gemini-2-5-pro): Gemini 2.5 Pro is a proprietary model from Google, released Jun 17, 2025. It ranks #42 of 351 on the LMArena text leaderboard with a rating of 1458. It costs $1.25 per million input tokens and $10 per million output tokens, with a 1M-token context window.
- [GPT-5.6 Sol](https://nexus.inc/models/gpt-5-6-sol): GPT-5.6 Sol is a proprietary model from OpenAI, released Jul 9, 2026. It ranks #43 of 351 on the LMArena text leaderboard with a rating of 1457. It costs $4 per million input tokens and $20 per million output tokens, with a 1.1M-token context window.
- [Kimi K2.6](https://nexus.inc/models/kimi-k2-6): Kimi K2.6 is an open-weight model from Moonshot AI, released Apr 20, 2026. It ranks #44 of 351 on the LMArena text leaderboard with a rating of 1455. It costs $0.65 per million input tokens and $3.41 per million output tokens, with a 256K-token context window.
- [Qwen3.7-Plus](https://nexus.inc/models/qwen3-7-plus): Qwen3.7-Plus is a proprietary model from Alibaba, released Jun 3, 2026. It ranks #45 of 351 on the LMArena text leaderboard with a rating of 1455. It costs $0.32 per million input tokens and $1.28 per million output tokens, with a 1M-token context window.

## Company

- [Pricing](https://nexus.inc/pricing)
- [Enterprise](https://nexus.inc/enterprise): Managed or Private deployment, SSO and SCIM, security, enterprise support
- [Blog](https://nexus.inc/blog)

## Optional

- [Nexus vs OpenRouter: choosing the right provider](https://nexus.inc/blog/nexus-vs-openrouter): Both put many models behind one API. They are built for different buyers.
- [Nexus vs LiteLLM: choosing the right provider](https://nexus.inc/blog/nexus-vs-litellm): Run it yourself, or own it without running it.
- [Auto Router v0.1.0: benchmarks](https://nexus.inc/blog/auto-router-v0-1-0-benchmarks): Score against cost for every model, and where Auto lands on the frontier.
- [Your keys and licenses. Free.](https://nexus.inc/blog/your-keys-and-licenses-free): Bring the API keys and AI plans you already pay for.
- [Managed or Private: running Nexus in the enterprise](https://nexus.inc/blog/managed-or-private): The same product either way. Choose who runs it and where your data lives.