Claude Instant 1.2: Faster and Smarter Text-Generating AI Model Released by Anthropic

Anthropic, the AI startup co-founded by ex-OpenAI execs, has released an updated version of its faster, cheaper, text-generating model available through an API, Claude Instant.

In its quest to create a “next-gen algorithm for AI self-teaching,” Anthropic has launched an improved version of Claude Instant, called Claude Instant 1.2. This updated model incorporates the strengths of Anthropic’s flagship model, Claude 2, and showcases significant improvements in areas such as math, coding, reasoning, and safety.

Enhanced Performance and Capabilities

In internal testing, Claude Instant 1.2 achieved remarkable scores of 58.7% and 86.7% on coding benchmarks and math questions respectively. These scores surpassed the performance of its predecessor, Claude Instant 1.1, which scored 52.8% on the coding benchmark and 80.9% on math questions.

The improvements are not limited to coding and math skills. Claude Instant 1.2 also excels in generating longer and more structured responses, better adherence to formatting instructions, quote extraction, multilingual capabilities, and question answering.

Enhanced Safety Features

Anthropic claims that Claude Instant 1.2 is less likely to generate incorrect or nonsensical text, a phenomenon known as “hallucination.” Additionally, the new version is more resistant to jailbreaking attempts, which involve bypassing the safety features set by the model’s creators. These enhancements fortify the reliability and integrity of the text generated by Claude Instant.

Larger Context Window and Extended Analytical Capacity

Claude Instant 1.2 boasts a context window equal in size to Claude 2’s context window – 100,000 tokens. The context window represents the text the model considers before generating additional text. As a result, both Claude Instant 1.2 and Claude 2 can analyze approximately 75,000 words, equivalent to the length of “The Great Gatsby.”

This increased context window helps prevent the model from “forgetting” recent conversations, enabling it to maintain continuity and coherence.

Competition and Ambitions

Claude Instant is Anthropic’s offering to compete with similar entry-level AI text-generation models from OpenAI, Cohere, AI21 Labs, and other startups. Despite being a robust and versatile model, Claude Instant is not the next-gen algorithm for AI self-teaching that Anthropic aims to develop.

Anthropic has already secured numerous customers and partners, including Quora, which provides access to Claude and Claude Instant through its generative AI app Poe. Claude also powers DuckDuckGo’s DuckAssist tool, offering direct answers to search queries in conjunction with OpenAI’s ChatGPT. Additionally, Claude serves as the technical backend for Notion AI, an AI writing assistant integrated with the Notion workspace.

Future Funding Requirements

Since its launch in 2021, Anthropic, led by former OpenAI VP of research Dario Amodei, has raised $1.45 billion in funding. However, the company estimates that it will need approximately $5 billion over the next two years to fully develop its envisioned chatbot and algorithm for AI self-teaching.

Conclusion

Anthropic’s updated Claude Instant 1.2 showcases improved performance, enhanced safety features, and extended capabilities. It represents a significant step towards Anthropic’s ultimate goal of developing a next-gen algorithm for AI self-teaching. With a growing list of customers and partners, Anthropic is poised to play a prominent role in the rapidly evolving field of AI text-generation.