Looking for an alternative to NanoGPT?

NanoGPT is a pay-as-you-go gateway to 650+ text and image models with a low-cost open-weight subscription; Standard Compute's plan also covers Claude, GPT and Gemini, which NanoGPT bills separately at list price.

TL;DR

Stay on NanoGPT if your agent does well on open-weight models. Switch if it needs Claude, GPT or Gemini for the hard steps.

NanoGPT raised its subscription price in May 2026

NanoGPT said nothing else about the subscription changed with the price, but that it "cannot rule out further changes in the future".

Date	Change
Feb 17, 2026	Weekly input-token cap, 10 concurrent requests and a 10-requests-per-10-seconds burst limit
Mar 1, 2026	New subscription signups paused
May 12, 2026	Price goes from $8 to $12 a month
Oct 13, 2026	Qwen 3.8 27B Uncensored no longer included in the subscription

From NanoGPT's blog and updates feed (sources at the end of the page). The February post called the $8 tier unprofitable "when people use it to the limit". The pricing page showed a Subscribe button again on October 13; we found no post saying when signups reopened.

What NanoGPT offers

Pay-as-you-go text at list price with no markup and no deposit fee when NanoGPT routes the provider; pinning a provider adds 5%, and your own keys cost 5% of the normal model price. PRO is $12 a month: 60 million input tokens a week on included models, 100 images a day and 5% off eligible paid text models. (Verified 2026-10-13.)

List-price pay-as-you-go with no markup and no deposit fee, by card
One of the largest catalogs, with image generation alongside text
Anthropic Messages, Responses and Chat Completions endpoints, with Claude Code, Codex and OpenCode docs
Teams with roles and per-member limits, and daily, weekly or monthly caps per API key
Opt-in providers that keep inference in the EU

Why people look for an alternative

Prices and the included lineup have both changed this year, and may change again
The subscription covers open-weight models only: on October 13 no Claude, GPT or Gemini model was included, and Kimi K3 was not either
The weekly token cap sounds large, but coding agents resend context on every step, and some models count double
Models leave the subscription over time

Which one fits your work

Pick	When
Standard Compute	Claude, GPT and Gemini are inside the plan budget, routed with open models, instead of billed separately at list price
	One monthly compute budget with no weekly window, from $19 a month
	The router picks a model per request from the lineup you select, so routine steps use efficient models
	Europe or US region rules on a key, team or company, rather than choosing EU providers one by one
NanoGPT	Your agent does well on open-weight models; for those, the subscription covers far more tokens per dollar
	You want pay-as-you-go at list price with no markup, from a $1 deposit
	You need a very large catalog, uncensored variants or image generation under one key
	Light use that a small prepaid balance covers

Switching takes one config change

Standard Compute is OpenAI-compatible, so any tool or SDK that lets you set a custom base URL migrates in minutes:

Base URL  = https://api.stdcmpt.com/v1
API key   = your Standard Compute key
Model     = standardcompute

Setup guides for OpenClaw, Hermes, OpenCode, Cursor, Cline, Aider and more are on the integrations page.

FAQ

Did NanoGPT raise its subscription price?

Yes, by half. NanoGPT announced the increase on May 5, 2026, citing more expensive models; the timeline above shows the earlier caps and the paused signups.

Does the NanoGPT subscription include Claude or GPT?

No. When we checked, every model marked as included in NanoGPT's live catalog was open-weight; Claude, GPT and Gemini are billed pay-as-you-go. Standard Compute plans include Claude, GPT and Gemini alongside open models within the monthly compute budget.

Can I use NanoGPT with Claude Code?

Yes. NanoGPT documents ANTHROPIC_BASE_URL=https://nano-gpt.com/api/v1 with your NanoGPT key. On Standard Compute you set ANTHROPIC_BASE_URL to https://api.stdcmpt.com; the same key serves Codex through /v1/responses and other agents through https://api.stdcmpt.com/v1.

Is NanoGPT cheaper than Standard Compute?

For open-weight models, usually: the subscription covers far more open-model tokens than our Starter plan's $20 budget. Standard Compute is the better fit when your agent needs frontier models for part of the work, since those are included in the plan and the router saves them for the steps that need them.

Sources: NanoGPT updates (May 5, 2026 price update); NanoGPT subscription update, February 2026; NanoGPT signups paused; NanoGPT pricing; NanoGPT llms-full.txt; NanoGPT model catalog API; NanoGPT Claude Code docs