Which model should I use?
Each lab now sells a family: a top model for hard thinking, a fast mid model for daily work and a small cheap one for volume. Start with the default.
The current families (October 2026):
| Lab | Most capable | Everyday | Fast or low-cost |
|---|---|---|---|
| OpenAI | GPT-6 Astra | GPT-6 Sol, GPT-6.1 Sol | GPT-6.1 Terra |
| Anthropic | Claude Opus 5.5 | Claude Sonnet 5.5 | Claude Haiku 5.5 |
| Gemini 4 Argon (Carbon is limited) | Gemini Flash | Gemma (small, open weights) | |
| xAI | Grok 4.7 | Grok 4.7 | Grok 4.7 Fast (quicker, costs more) |
Practical advice:
- For chat and everyday work, use the default your app picks. They are very good.
- Switch to the top model for complex analysis, long documents or hard reasoning.
- When building tools that run thousands of times, use the cheapest model that passes your tests.
- The lead changes every few months. Do not lock your company into one model; keep your prompts, files and processes portable.
Prices fall fast: Anthropic cut Opus 5.5 running costs by about 40% versus Opus 5, and OpenAI priced Terra 50% below its previous Terra model.
Sources
- ChatGPT Work and Codex (Help Center) OpenAI, Oct 2026
- Introducing GPT-6.1 Terra OpenAI, 26 Oct 2026
- Introducing Claude Opus 5.5 Anthropic, 22 Sep 2026
- Introducing Claude Haiku 5.5 Anthropic, 7 Oct 2026
- Gemini 4 Argon is now available for everyone Google, 13 Oct 2026
- Introducing Gemini 4 Carbon Google, 28 Oct 2026
- Gemini at Work 2026: Introducing Gemini agent Google Cloud, 8 Oct 2026
- xAI API release notes xAI, updated 27 Oct 2026
Learn it hands on, with us.
Free AI build nights in Kuala Lumpur. Bring one real problem and leave with a working tool.