Releasing Claude Instant 1.2

Businesses working with Claude can now access our latest version of Claude Instant, version 1.2, available through our API. Claude Instant is our faster, lower-priced yet still very capable model, which can handle a range of tasks including casual dialogue, text analysis, summarization, and document comprehension.

Claude Instant 1.2 incorporates the strengths of our latest model Claude 2 in real-world use cases and shows significant gains in key areas like math, coding, reasoning, and safety. It generates longer, more structured responses and follows formatting instructions better. Instant 1.2 also shows improvements in quote extraction, multilingual capabilities, and question answering.

Claude Instant 1.2 outperforms Claude Instant 1.1 on math and coding, achieving 58.7% on the Codex evaluation compared to 52.8% in our previous model. It also scored 86.7% on the GSM8K benchmark, compared to 80.9% for Claude Instant 1.1.

Performance of Claude Instant 1.1 compared to 1.2

Our latest model has also improved on safety. It hallucinates less and is more resistant to jailbreaks, as shown in our automated red-teaming evaluation.

Safety evaluation of Claude models. Lower is better.

Developers looking to work with Claude Instant 1.2 can now call our latest model over our API (pricing can be found here). If you’re a business and you’d like to work with us, you can indicate your interest here.