text.chat4 providers
DeepSeek V4 Flash 0731
DeepSeek · open weights
86%below what the other 3 providers typically charge
On DeepInfra$0.090 per 1M
call it asdeepseek-ai/DeepSeek-V4-Flash-0731
Live catalog, refreshed every day
The right AI model, on the providers you already use
Tell it what you need and which providers you have accounts with. It answers with the best-suited model, named exactly as your provider's API expects it, at the cheapest price you can get it.
Free for 1,000 calls a month. No card, no SDK: plain HTTPS.
GET /v1/resolve200 OK
You ask
- task=text.chat
- budget=cheap
- providers=gemini,openrouter
It answers
- model
- google/gemini-3.5-flash-lite
- call it on
- gemini · gemini-3.5-flash-lite
- typical call
- $0.00122 · per 1M in $0.30 / out $2.50
- also on
- openrouter $0.00129 · +5.5% credit fee
- reason
- best text.chat model in the cheap band on your providers
Answers for the providers you already have
- OpenAI
- Anthropic
- Google Gemini API
- Google Vertex AI
- Amazon Bedrock
- Azure OpenAI
- Mistral AI
- DeepSeek
- xAI
- Cohere
- Groq
- OpenRouter
- Fireworks AI
- Together AI
- DeepInfra
- Cerebras
- Ollama
- LM Studio
- llama.cpp
- MLX
Live API · no account needed
Try it live
Ask the real API one of these, and see the model it picks, where to call it and what a call costs. Sandbox calls aren't tied to an account.
Waiting for a question
Pick a question to ask the live API.
Deals of the day
Same model, a much smaller bill
Models sold by three or more providers, where one of them charges far less than the rest. Recomputed after every catalog refresh.
Catalog of
See all 26 dealsCompare all 4 prices
text.chat8 providers
GLM-5.3-Flash
Zhipu AI (Z.ai) · open weights
78%below what the other 7 providers typically charge
On OpenRouter, fee included$0.052 per 1M
call it asz-ai/glm-5.3-flash
text.chat5 providers
gpt-oss-20b
OpenAI · open weights
69%below what the other 4 providers typically charge
On Experiential Labs$0.018 per 1M
call it asgpt-oss-20b
- 78%GLM-5.3-FlashOpenRouter, 8 providers$0.052 vs typical $0.237
- 69%gpt-oss-20bExperiential Labs, 5 providers$0.018 vs typical $0.058
A deal is a metered, generally available price on a model at least three providers sell, measured against what the others typically charge, fees included. Trial, free and unverified prices never count.
Why it exists
How it compares
Other tools pick one model for everyone, or route each request through their own account. This one answers for your providers and stays out of the request path.
One model, not a menu
Leaderboards and catalogs list hundreds of models. You describe the need (a task or a purpose, the capabilities it requires, a budget) and get one answer, with the reason and the next alternatives.
The cheapest provider for it
The same model is often sold by its maker, a cloud platform and a router at different prices and fees. You get the cheapest offering that has every capability you asked for, and the others as fallbacks.
Your providers only
You list the providers you have accounts with. A model you can't call is never proposed, and the answer names it exactly as that provider's API expects it. Your provider keys never leave your servers.
Three steps
How it works
Create a free account and an API key.
RHUMB_KEY=nnm_test_…Call /v1/resolve from your server with your providers and what you need.
GET /v1/resolve?task=…&providers=…Cache the answer for a day and call the model it names, with a fallback in config.
model.provider_model_id
Pricing
Start free, grow when it pays
Start free with 1,000 calls a month. Paid plans from $9 a month.
Builder
$9 / month
25,000 calls a month
Pro
$39 / month
250,000 calls a month
Scale
$149 / month
2,000,000 calls a month
Stop picking models by hand
Ask once, cache the answer, and let tomorrow's catalog pick tomorrow's model.