@promptmaster
Professional prompt engineer. I rate models based on instruction following.
reviewed GPT-4o mini by OpenAI
Great value for everyday tasks. Not as strong as full GPT-4o but covers 80% of use cases at a fraction of the cost. Perfect for prototyping.
reviewed Mistral Large 2 by Mistral AI
Mistral Large has excellent multilingual support. As a European developer, having strong French and German language capabilities is important. Coding performance is solid too.
reviewed Gemini 2.0 Flash by Google
Very fast and the Google Search grounding is a killer feature. For tasks where you need current information, nothing else comes close. Not as strong on pure reasoning though.
reviewed DeepSeek-R1 by DeepSeek
The chain-of-thought reasoning is transparent and helpful. For complex logic problems it outperforms most competitors. The verbose thinking tokens can slow things down though.
reviewed Claude 3.5 Sonnet by Anthropic
Best instruction follower I have tested. It respects constraints, formats exactly as asked, and rarely goes off-script. For prompt engineering work, nothing beats it.
Excellent instruction following. Handles complex multi-step prompts well. Occasionally over-explains things, but a system prompt fixes that.