The newest rows
Three open-weight models were added this week. SenseNova-Loop-9B loops 32 decoder layers into 56 executions and takes text and images over a 512K context. Maxwell V1.0 Flash is a 35B-total, 3B-active sparse mixture-of-experts model with a 256K context. Rumble-2-Flash is a 0.5B speech-recognition model for 45 languages.
All three are listed as Coming soon. Their weights are public under Apache 2.0, but no inference API serves them with a price we can publish. AnyRouter lists them now so they appear on /models, with a clear note instead of a fake route. We will add a hop and a price as soon as a host serves them.
The GPT-6.2 Astra and Sol rows are already routable. This week added free and paid Respan routes for Sol, a Singapore QwenCloud route that also speaks the Responses API, and Respan BYOK for both models.
Why one key reaches so many models
A model gets one catalog id, then as many upstream hops as serve its weights or API. The router tries those hops in order and the circuit breaker skips the ones that fail. Adding a provider does not create a second model row or a new allowlist: the same id keeps working with whichever key you have.
Most of those hops are BYOK. You save a key under Dashboard, BYOK, the request goes out under your own account, and your provider bills you directly. AnyRouter adds no per-token charge on a BYOK hop. That is why a small team can offer more than a hundred provider routes without holding a funded account at every one of them.
- Direct model labs: OpenAI, Anthropic, Google, xAI, DeepSeek, Qwen, Mistral AI, Microsoft.
- Unified gateways: OpenRouter, Vercel AI Gateway, Requesty, ZenMux, Llamport, AIHubMix, NanoGPT.
- Specialist hosts: Respan, Chutes, Infron, CrowLLM, SubQ, Liquid AI, FPT Cloud, QwenCloud, SenseTime.
A free SKU is a hop on the same row, not a separate listing. The 24-hour free GPT-6.2 Sol wire from Respan, Infron's free DeepSeek-V4-Flash-0731 wire and CrowLLM's free wire all sit on their canonical model pages and can be filtered with ?f=free.
Try a model with your own key
Add a provider key once, then change only the model field to switch between models and upstreams.
curl https://anyrouter.dev/api/v1/chat/completions \
-H "Authorization: Bearer $ANYROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-6.2-sol",
"messages": [{"role": "user", "content": "Which is larger, 9.9 or 9.11?"}]
}'The model catalog shows which access each row supports. A BYOK tag means the route is yours to supply; a Coming soon tag means the weights exist but no API does yet.
Playbooks: AI SDK, coding agents and OpenAI SDK.
Related posts
- Run coding agents through one gatewayGuidesJuly 12, 2026 · Point Claude Code, Codex, OpenCode, Cursor, or any OpenAI-compatible tool at AnyRouter and get one key for every model, unified billing, per-request audit logs, and app attribution. Two environment variables, or one npx command with the anyr CLI launcher.
- Use AnyRouter with the Vercel AI SDKGuidesJuly 19, 2026 · @anyr/ai-sdk-provider wraps the AnyRouter gateway as a standard Vercel AI SDK provider. Call createAnyRouter(), pass a provider/model id, and generateText, streamText, and generateObject work unchanged, switching models is a one-string edit.
- What is AnyRouter? A plain-English guide to AI model routingGuidesAugust 5, 2026 · An AI model router lets you call many LLMs through one API. Here's what AnyRouter is, why it exists, and whether it's useful for your next project.
Route your first request in 2 minutes
Start free with your own keys, or top up and pay per token. Get $4/mo in credits and free models on Go — $2/mo, or free when you donate a provider key.