LLM names are confusing.
Today’s latest LLM models: GPT 5.5 (OpenAI), Claude Opus 4.7 (Anthropic), Gemini 3.1 Pro (Google).
The prefixes (GPT, Claude Opus, Gemini) are pretty random. They don’t mean much other than a name. The things that follow matter more.
Numbering (e.g. 5.5) usually indicates “version”, and higher number means more recent, which is general means stronger model (e.g. GPT 5.4 precedes GPT 5.5).
Some companies have “model series”, e.g. Claude models, “Haiku” is smallest/cheap, “Sonnet” is medium, “Opus” is most intelligent/expensive. For each series there is also version, e.g. there are Sonnet 4.5, Sonnet 4.6, Opus 4.5, Opus 4.6 etc.
In general, within the same version, bigger models are more intelligent. E.g. Opus 4.6 > Sonnet 4.6. Similarly for within the same serie: Opus 4.6 > Opus 4.5.
To make things more confusing, LLM model naming is pretty random. While the above are typical conventions, LLM providers do not have to follow them. E.g. OpenAI went from GPT-4.5 directly to GPT-5. There was no GPT-4.6.
Cheatsheet
| LLM Provider | Models | Notes |
|---|---|---|
| OpenAI | GPT-5.5 Pro, GPT-5.5, GPT-5.4, GPT-5.3 | Larger version number is newer LLM. Pro is a dedicated series that thinks for much longer. |
| Anthropic | Claude Opus 4.7, Claude Sonnet 4.6, Claude Haiku 4.5 | Intelligence: Opus > Sonnet > Haiku Note: at the time of writing, Anthropic only released up to Sonnet 4.6 and Haiku 4.5. Opus 4.7 released does not mean there is also Sonnet 4.7 or Haiku 4.7. |
| Gemini 3.1 Pro, Gemini 3 Flash | Google has two model series: Pro and Flash. Intelligence: Pro > Flash. |
Deepseek currently has two LLMs: DeepSeek-V4-Flash and DeepSeek-V4-Pro. Which part of the name is the series name and which part is the version number? Which one would you expect to be the more intelligent model?