Compare two AI models
Pick any two and see what each one costs, how much it holds and when it shipped. The numbers are the labs' own, and the monthly figures are worked out from them.
58 models · updated 2026-09-05 12:15 UTC
| GPT-6 Astra | Gemini 3.8 Flash | |
|---|---|---|
| Lab | OpenAI | |
| Released | 2026-09-03 | 2026-09-02 |
| Status | released | released |
| Built in | US | US |
| Context window | 1M | 1M |
| Longest reply | 128K | 65.5K |
| Input, per million | $10.00 | $0.75 |
| Output, per million | $50.00 | $3.75 |
| API name | gpt-6-astra | gemini-3.8-flash |
| A few questions a day, a month | $4.80 | $0.36 |
| Steady daily use, a month | $26 | $1.92 |
| Something running all day, a month | $600 | $45 |
The ones people ask about
- GPT-6 Astra vs Gemini 3.8 FlashOpenAI against Google
- GPT-6 Astra vs Claude Fable 5.1OpenAI against Anthropic
- GPT-6 Astra vs Grok 4.6OpenAI against xAI
- GPT-6 Astra vs DeepSeek-V4-ProOpenAI against DeepSeek
- Gemini 3.8 Flash vs Claude Fable 5.1Google against Anthropic
- Gemini 3.8 Flash vs Grok 4.6Google against xAI
- Gemini 3.8 Flash vs DeepSeek-V4-ProGoogle against DeepSeek
- Claude Fable 5.1 vs Grok 4.6Anthropic against xAI
- Claude Fable 5.1 vs DeepSeek-V4-ProAnthropic against DeepSeek
- Grok 4.6 vs DeepSeek-V4-ProxAI against DeepSeek
- GPT-6 Astra vs GPT-5.6 LunaOpenAI against OpenAI
- Gemini 3.8 Flash vs Gemini 3.1 Flash-LiteGoogle against Google
- Claude Fable 5.1 vs Claude Haiku 4.5Anthropic against Anthropic
- Grok 4.6 vs Grok Build 0.1xAI against xAI
- DeepSeek-V4-Pro vs DeepSeek-V4-FlashDeepSeek against DeepSeek
- Mistral Medium 3.5 vs Ministral 3 14BMistral against Mistral
Or by brand
Whole line-ups rather than single models, for when the question is which company rather than which version.
- ChatGPT vs ClaudeOpenAI against Anthropic
- ChatGPT vs GeminiOpenAI against Google
- Claude vs GeminiAnthropic against Google
What we compare, and what we don't
Price, context window, longest reply, release date and status. All of it published by the lab, all of it checkable from the model's own page.
We don't score models against each other. Benchmarks move every month and none of them predict how a model handles your work. The price table is the honest half of the question, and it's the half that decides most of the bill.