All AI models on Freepay
50 models from nine companies in one chat. No subscription, you pay only for what you ask.
Prices in the tables: your text / answer per 1M tokens at the maker. On Freepay every model is 10x cheaper: a key costs from $0.18 per 1M balance tokens, and the “Freepay rate” column shows how much a model charges compared with Claude Sonnet 5.5. All prices in one table on the Pricing page.
Claude: models by Anthropic
Claude is made by Anthropic, a US company. People pick it for natural writing, strong code and the way it handles long documents. On claude.ai you need an account and a phone number, the free plan has tight message limits, and Claude Pro is a monthly subscription.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Claude Fable 5.2 | The strongest Claude. For the hardest work: big code, deep analysis, long projects. | $8 / $40 | ×4 / ×20 |
| Claude Fable 5.1 | The previous Fable, the strongest Claude. The new Fable 5.2 is smarter and cheaper, but this one still works. | $10 / $50 | ×5 / ×25 |
| Claude Fable 5 | The first Fable. Powerful, but pricier than the newer versions: pick it if you are used to it. | $10 / $50 | ×5 / ×25 |
| Claude Opus 5.5 | Claude’s flagship for hard writing, code and analysis. Half the price of Fable and handles most difficult tasks. | $4 / $20 | ×2 / ×10 |
| Claude Opus 5 | Last-generation Opus. Strong, but the new Opus 5.5 is cheaper and smarter. | $5 / $25 | ×2.5 / ×12.5 |
| Claude Opus 4.8 | The last Opus of the fourth generation. Reliable for hard writing and code. | $5 / $25 | ×2.5 / ×12.5 |
| Claude Opus 4.7 | Opus 4.7: a strong last-generation model for hard tasks. | $5 / $25 | ×2.5 / ×12.5 |
| Claude Opus 4.6 | Opus 4.6: a proven last-generation model for writing and code. | $5 / $25 | ×2.5 / ×12.5 |
| Claude Opus 4.5 | Opus 4.5: one of the models that made Claude popular with programmers. | $5 / $25 | ×2.5 / ×12.5 |
| Claude Sonnet 5.5 | Claude’s everyday model. Natural writing and good code, the best value of the line. | $2 / $10 | ×1 / ×5 |
| Claude Sonnet 5 | The previous Sonnet. Same price as Sonnet 5.5, but the new version is smarter. | $2 / $10 | ×1 / ×5 |
| Claude Sonnet 4.6 | Fourth-generation Sonnet: good at writing and code, but pricier than the new versions. | $3 / $15 | ×1.5 / ×7.5 |
| Claude Sonnet 4.5 | Sonnet 4.5: a proven last-generation model for writing and code. | $3 / $15 | ×1.5 / ×7.5 |
| Claude Haiku 5.5 | The fastest and cheapest Claude. For short answers, translations and simple questions. | $0.1 / $0.5 | ×0.05 / ×0.25 |
| Claude Haiku 4.5 | Last-generation fast Claude. The new Haiku 5.5 is ten times cheaper. | $1 / $5 | ×0.5 / ×2.5 |
ChatGPT: models by OpenAI
GPT models are made by OpenAI and power ChatGPT, the best-known AI chat. It answers any question, does maths, explains and helps with study. ChatGPT needs an account, the free plan limits the best models, and ChatGPT Plus is a monthly subscription.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| GPT-6.1 Sol | ChatGPT’s everyday model. Answers any question, writes text and code. | $2 / $10 | ×1 / ×5 |
| GPT-6 Astra | The smartest ChatGPT model. For hard problems, research and difficult maths. | $10 / $50 | ×5 / ×25 |
| GPT-6.1 Terra | The mid-size ChatGPT: fast and inexpensive, enough for most questions. | $1 / $5 | ×0.5 / ×2.5 |
| GPT-6 Luna | The cheapest sixth-generation ChatGPT. For simple questions and short answers. | $0.1 / $0.5 | ×0.05 / ×0.25 |
| GPT-6 Sol | The previous Sol. Same price as 6.1 Sol, the new version is a bit stronger. | $2 / $10 | ×1 / ×5 |
| GPT-5.6 Sol | Fifth-generation Sol: still a good ChatGPT for writing and questions. | $2 / $10 | ×1 / ×5 |
| GPT-5.6 Terra | The mid-size fifth-generation ChatGPT: fast and inexpensive. | $1 / $5 | ×0.5 / ×2.5 |
| GPT-5.6 Luna | The cheapest fifth-generation ChatGPT for simple questions. | $0.1 / $0.5 | ×0.05 / ×0.25 |
| GPT-5.4 Mini | A compact ChatGPT for short answers and simple tasks. | $0.75 / $4.5 | ×0.375 / ×2.25 |
| GPT-5.4 Nano | A tiny ChatGPT: very cheap, for the simplest questions. | $0.2 / $1.25 | ×0.1 / ×0.625 |
Gemini: models by Google
Gemini is Google’s AI. It answers fast, works well with large documents and knows a lot of facts. Gemini needs a Google account, and the stronger models come with a paid Google AI plan.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Gemini 4 Argon | The strongest Gemini. For large documents, analysis and hard questions. | $2 / $12 | ×1 / ×6 |
| Gemini 3.1 Pro | Gemini Pro: works through long texts, spreadsheets and documents. | $2 / $12 | ×1 / ×6 |
| Gemini 3 Pro | Gemini 3 Pro: the previous Pro for documents and analysis. | $2 / $12 | ×1 / ×6 |
| Gemini 2.5 Pro | Gemini 2.5 Pro: the senior last-generation model, inexpensive for its level. | $1.25 / $10 | ×0.625 / ×5 |
| Gemini 3 Flash | A fast Gemini for short answers and summaries. | $0.5 / $3 | ×0.25 / ×1.5 |
| Gemini 3.6 Flash | The newest fast Gemini: answers in seconds and costs little. | $0.5 / $3 | ×0.25 / ×1.5 |
| Gemini 3.5 Flash | Fast Gemini 3.5 for everyday questions. | $0.5 / $3 | ×0.25 / ×1.5 |
| Gemini 2.5 Flash | A last-generation fast Gemini, even cheaper than the new Flash models. | $0.3 / $2.5 | ×0.15 / ×1.25 |
| Gemini 2.5 Flash Lite | The cheapest Gemini. For simple questions and lots of requests. | $0.1 / $0.4 | ×0.05 / ×0.2 |
Grok: models by xAI
Grok is made by xAI, Elon Musk’s company. It answers bluntly, reasons well and writes code. Grok lives in X and on grok.com, needs an account, and its best models come with a paid plan.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Grok 4.7 | The newest Grok: straight answers, reasoning and analysis. | $2 / $6 | ×1 / ×3 |
| Grok 4.6 | Grok 4.6: the previous version, same price as 4.7. | $2 / $6 | ×1 / ×3 |
| Grok 4.5 | Grok 4.5: a strong previous version. | $2 / $6 | ×1 / ×3 |
| Grok 4.3 | Grok 4.3: cheaper than the new versions, for everyday questions. | $1.25 / $2.5 | ×0.625 / ×1.25 |
| Grok Build 0.1 | Grok for programming: writes and fixes code. | $1 / $2 | ×0.5 / ×1 |
DeepSeek: models by DeepSeek
DeepSeek is a Chinese model that became famous for nearly matching ChatGPT at a fraction of the price. It is strong in maths, logic and code. DeepSeek’s own chat is free but needs an account and is often overloaded at peak hours.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| DeepSeek Flash | A fast and very cheap DeepSeek. For study, questions and lots of requests. | $0.3 / $1.2 | ×0.15 / ×0.6 |
| DeepSeek V4 Pro | A strong DeepSeek for maths, logic and code at a low price. | $1.32 / $3.96 | ×0.66 / ×1.98 |
GLM: models by Zhipu AI (Z.ai)
GLM is made by Zhipu AI, a Chinese company. It is a fast, inexpensive model for everyday tasks and code. The maker’s site needs an account and its paid plans are billed by card.
Kimi: models by Moonshot AI
Kimi is made by Moonshot AI in China. It keeps long conversations and big tasks on track, and the Thinking version reasons before it answers. Kimi’s site needs an account and works mostly in Chinese and English.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Kimi K2 | Kimi K2: long conversations and multi-step tasks. | $0.6 / $2.5 | ×0.3 / ×1.25 |
| Kimi K2 Thinking | Kimi K2 that reasons first and answers second. For maths, logic and plans. | $0.6 / $2.5 | ×0.3 / ×1.25 |
Qwen: models by Alibaba Cloud
Qwen is Alibaba’s AI. It writes good code, translates well and knows many languages. Qwen is also on Alibaba’s own site, which needs an account.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Qwen3 Max | The strongest Qwen: general questions, writing and translation. | $1.2 / $6 | ×0.6 / ×3 |
| Qwen3 Coder | Qwen for programming: writes and explains code. | $1 / $5 | ×0.5 / ×2.5 |
Mistral: models by Mistral AI
Mistral is a French AI. It knows European languages well, answers fast and costs little. Mistral’s Le Chat needs an account, and its paid plan is a monthly subscription.
| Model | Good for | Official, $ per 1M | Freepay rate |
|---|---|---|---|
| Mistral Large | Mistral’s flagship: writing and reasoning, especially in European languages. | $2 / $6 | ×1 / ×3 |
| Mistral Medium | An inexpensive Mistral for everyday questions. | $0.4 / $2 | ×0.2 / ×1 |
| Codestral | Mistral’s model for programming. | $0.3 / $0.9 | ×0.15 / ×0.45 |
Frequently asked questions
Which AI model should I pick?
For writing and code take Claude Sonnet 5.5, for any question and study GPT-6.1 Sol, for large documents Gemini 3.1 Pro, and if price matters, DeepSeek Flash.
Can I switch between models?
Yes. You pick the model right in the input field, and the balance is shared.
Is there a subscription?
No. You buy a key once, and the balance does not expire.
Is there a free model?
Yes. In the free chat KLINCH AI answers with no sign-up and no key. The other models work with a key.