onomeoonomeoBeta

Models

Top up

Free-model allowance

About 80% left

Shared by all accounts; what is used comes back gradually over 5 hours

All models

Last 7 days

48 models

How these figures are worked out

Updated Nov 22, 2026, 03:16 UTCAvailability is the share of all calls made through the site over the last 7 days, scheduled tests included, that got a reply. Response time averages the successful scheduled tests.Big models first, lowest price first; then the other models, by availability, then response time

Compare#ModelAvailabilityResponseReference price per million output tokensFree left
1claude-haiku-5.5Anthropic100%—avg.—Chat
2gpt-6-lunaOpenAI99%1.3savg.—Chat
3deepseek-v4.1-flashDeepSeek98%2.3savg.—Chat
4kimi-k3Moonshot AI100%—avg.—Chat
—gpt-6.1-terraOpenAINo data yet—avg.—Chat
—kimi-k3.1-proMoonshot AINo data yet—avg.—Chat
—claude-opus-5.5AnthropicNo data yet—avg.—Chat
—claude-sonnet-5.5AnthropicNo data yet—avg.—Chat
—gpt-6-astraOpenAINo data yet—avg.—Chat
—grok-4.7xAINo data yet—avg.—Chat
—gpt-6.1-solOpenAINo data yet—avg.—Chat
—claude-fable-5.1AnthropicNo data yet—avg.—Chat
—mistral-large-4Mistral AINo data yet—avg.—Chat
—qwen3.8-maxQwenNo data yet—avg.—Chat
—muse-wonder-1.0MetaNo data yet—avg.—Chat
—gemini-4-argonGoogleNo data yet—avg.—Chat
—claude-fable-5.2NewAnthropicNo data yet—avg.—Chat
—gpt-5.6-lunaOpenAINo data yet—avg.—Chat
—glm-5.3-flashZhipu AINo data yet—avg.—Chat
—gemini-3.8-flashGoogleNo data yet—avg.—Chat
—mimo-v2.6-flashNewXiaomiNo data yet—avg.—Chat
—mimo-v2.6-proNewXiaomiNo data yet—avg.—Chat
—hy4-previewTencentNo data yet—avg.—Chat
—deepseek-v4-proDeepSeekNo data yet—avg.—Chat
—mistral-small-2603NewMistral AINo data yet—avg.—Chat
—gemini-4-carbonNewGoogleNo data yet—avg.—Chat
—gemini-4-hydrogenNewGoogleNo data yet—avg.—Chat
—glm-5v-turboNewZhipu AINo data yet—avg.—Chat
—grok-4.20NewxAINo data yet—avg.—Chat
—grok-build-0.1NewxAINo data yet—avg.—Chat
—kimi-k2.7-codeNewMoonshot AINo data yet—avg.—Chat
—mistral-medium-3-5NewMistral AINo data yet—avg.—Chat
—muse-spark-openNewMetaNo data yet—avg.—Chat
—qwen3.8-2.4t-a95bNewQwenNo data yet—avg.—Chat
—qwen3.8-flashNewQwenNo data yet—avg.—Chat
—seed-2-1-turboNewByteDanceNo data yet—avg.—Chat
—gpt-6.2-solNewOpenAINo data yet—avg.—Chat
—gpt-6.2-astraNewOpenAINo data yet—avg.—Chat
—gemini-4-neonNewGoogleNo data yet—avg.—Chat
—qwen4-maxNewQwenNo data yet—avg.—Chat
—hy4NewTencentNo data yet—avg.—Chat
—hy4-flashNewTencentNo data yet—avg.—Chat
5gemma-4-31bNewGoogle95%1.8savg.Free99% leftChat
6glm-4.7-flashZhipu AI95%14savg.Free—Chat
7glm-5.3Zhipu AI91%2.2savg.Free—Chat
8mistral-small-3.1NewMistral AI91%11savg.FreeUsed upChat
9qwen3.8-27bQwen87%4.4savg.FreeUsed upChat
10llama-4-scoutNewMeta86%5.0savg.FreeUsed upChat
—autoRecommended: picks for you——avg.Free—Chat

These are reference unit prices. Paid text models are charged by actual usage, including input, output and reasoning. Cache reads use the cache rate; supported cache writes have separate prices. A request is never charged more than the most it was estimated to cost when sent.

This week's model changes

This week's changes

Removed
gemini-4-neon-freeglm-5.2gpt-oss-120b

No successful replies

Every model answered at least once this week.

More free LLM APIs are compared in the Free LLM API guide by Dmytro Klymentiev.

Pricing

Chat and video are charged by usage; images are charged per picture.

Free

$0

  • Free models: 30 calls every 5 hours
  • Chat in the browser, with 3 free tries of Claude, GPT and other big models
  • Pay $1 by card, get $5 of credit

Free models from OpenAI, Google, DeepSeek and more

Pay as you go

One payment. What you pay becomes your balance.

Amount

Through December 1 (UTC), one top-up of $10 or more adds 20% extra to your balance.

Your first top-up adds $2 extra balance, once per account; with the time-limited bonus you get whichever is more, not both.

  • Big modelsUnlocks every big modelIncluding GPT-6.2 Astra, Gemini 4 Neon, GPT-6.2 Sol, Qwen4 Max and more, charged to the balance by actual usage.A big model in a coding tool costs much more; the first large call shows its estimated cost first.
  • Premium picturesAbout166pictures from GPT Image 2.6About 83 from Nano Banana 2.1, 55 from Nano Banana Pro
  • VideoOpens with a single top-up of $10 or moreMade from text, with sound

Each line is an estimate for that use alone; all draw on one balance.

  • The balance never expires
  • No automatic renewal
  • For 35 days after payment: 90 free-model calls every 5 hours
  • With $10 or more paid in total and a balance left: 300 free-model calls every 5 hours, free-model requests of up to 128,000 characters, and access when the site-wide free limit is reached

PayPal payments go through Ko-fi. Use the email you sign in with here.

Payments are not refunded. See the Terms of Use.

Billing rules

Paid models are charged by usage; rewards apply to eligible web answers

  • These are reference unit prices. Paid text models are charged by actual usage, including input, output and reasoning. Cache reads use the cache rate; supported cache writes have separate prices. A request is never charged more than the most it was estimated to cost when sent.
  • Output usage includes model reasoning. A request whose reasoning uses up the output limit before an answer is written is charged for the tokens used; other requests with no usable delivered content (a provider error, a blank reply) are not charged.
  • Images read by text models are included in input tokens, with no additional per-image fee.
  • On big models, the text read from a file you attach is charged like the rest of the message, each time it goes with a question.
  • Sign in with an email code, Google, GitHub or Discord. No password and no card needed.
  • Claude, GPT and other big models use your top-up balance; earned rewards can cover selected models in web chat; an account that has not topped up can try them free 3 times in the web chat.
  • Top up from $5. Your balance never expires, and nothing renews on its own.
  • Up to 12 requests a minute per key.
  • Free models: up to 30 calls in each 5-hour window per account, or 90 for 35 days after a payment. A long message counts as several calls: one for every 8,000 counted characters, and one more for any part left over. Characters are counted across the whole conversation sent, earlier messages included; each Chinese character counts as one, every 4 other characters as one.
  • Accounts with $10 or more paid in total and a balance left get 300 free-model calls every 5 hours, requests of up to 128,000 characters, and access when the site-wide free limit is reached.

Where the models come from The current catalogue shows available models and prices. Paid calls use your balance, while earned rewards can cover eligible answers in browser chat. Account and rate limits still apply.Read the FAQ

Buying

Select Top up. Waffo Pancake, our merchant of record, processes the payment. When you are signed in, the balance is added to that account; you can also pay without signing in and receive the balance later by signing in or registering with the email you paid with. See the Terms of Use for details.

Frequently asked questions

What is the balance?

Each payment adds to your balance the amount the top-up page states, shown in dollars. Claude, GPT and other big models are charged to it by usage; so is DeepSeek for accounts that have topped up, without counting toward free-model limits. Free models do not touch the balance; they are limited by calls. The balance never expires and nothing renews automatically; top up again whenever it runs out.

How much does one reply cost?

These are reference unit prices. Paid text models are charged by actual usage, including input, output and reasoning. Cache reads use the cache rate; supported cache writes have separate prices. A request is never charged more than the most it was estimated to cost when sent. On send, your balance is checked against the most the request could cost; when it ends, only actual usage is charged.

onomeo is an independent product and is not affiliated with any model provider.