Stackness

LLMs tools

Most used tools

  1. #1

    Anthropic's most capable Claude family, for deep reasoning and long agentic runs

    1 user
  2. #2

    Anthropic's balanced Claude family, tuned for everyday coding and analysis

    1 user
  3. #3

    Anthropic's fastest Claude family, for high-volume and latency-sensitive work

    1 user
  4. #4

    Anthropic's Claude family offered alongside Opus, Sonnet and Haiku

    1 user
  5. #5

    Mistral's flagship family for reasoning, coding and multilingual work

    1 user
  6. #6

    OpenAI's general-purpose model family behind ChatGPT and the OpenAI API

    0 users
  7. #7

    Google's higher-capability Gemini family for complex reasoning and long context

    0 users
  8. #8

    Google's fast, low-cost Gemini family for high-throughput work

    0 users
  9. #9

    xAI's model family, served through the Grok apps and the xAI API

    0 users
  10. #10

    Cohere's enterprise family tuned for retrieval-augmented generation and tool use

    0 users
  11. #11

    Amazon's foundation model family served through Amazon Bedrock

    0 users
  12. #12

    LLM inference engine in C/C++ for running large language models locally.

    0 users
  13. #13

    A high-performance language model by DeepSeek demonstrating advanced reasoning and coding capabilities.

    0 users
  14. #14

    Large language model by DeepSeek optimized for reasoning and code generation.

    0 users
  15. #15

    Large language model providing frontier intelligence capabilities.

    0 users
  16. #16

    Large language model by MiniMax offering advanced inference capabilities.

    0 users
  17. #17

    High-throughput and memory-efficient inference engine for large language models.

    0 users

Moves in LLMs

by StacknessLLMsbackdated

Deciding what a model should see and in what order: which files, which history, which tools. The successor argument to prompt engineering, once the context window stopped being the scarce part.

63

Fetching the relevant documents at question time and putting them in the model's context, instead of relying on what the weights happen to remember.

63

by StacknessLLMsbackdated

Treating the wording of a model's instructions as an engineering artifact worth versioning, testing and reviewing rather than something typed once and forgotten.

63

Adoption history

Claude Opus

Claude Sonnet

Claude Haiku