AI Based Code Editor

AI Integrated in IDE

Paling enak dan kaya fitur: https://cursor.sh/

Dapat pendanaan USD 8 M dari Open AI di Oktober 2023

Code Story: https://codestory.ai/

Support local LLM: https://docs.codestory.ai/local-models

Dapat pendanaan sekian dari YCombinator (Agustus 2023), USD 500K

AI as VS Code Extension

Codeium: https://codeium.com/

Rating 4.9/5 di VSCode Marketplace, https://marketplace.visualstudio.com/items?itemName=Codeium.codeium

Dapat pendanaan USD 65 M di Januari 2024

Support local LLM seperti: Codellama, Starcoder, DeepSeek Coder, WizardCoder

Cara pakai local LLM: https://sourcegraph.com/blog/local-code-completion-with-ollama-and-cody

Per Juni 2023 sudah dapat funding USD 248 M

Sourcegraph Cody: https://sourcegraph.com/

LAIN-LAINNYA

https://pieces.app/
- https://docs.pieces.app/features/pieces-copilot

Local LLM VS Code Extension

Cara pakai lihat di https://medium.com/@smfraser/how-to-use-a-local-llm-as-a-free-coding-copilot-in-vs-code-6dffc053369d

Per Maret 2024 bisa gunakan DeepSeek Coder 33B, terbaik sementara menurut data di https://evalplus.github.io/leaderboard.html

Begini cara installnya di Ollama, dan Ollama sendiri sudah support Open AI compatibility lho, jadi bisa run DeepSeek Coder dari Cursor karena cursor mendukung Open AI compatible

Hosted AI LLM Murah dan Bagus

Mau murah dan performa bagus? Gunakan open source LLM untuk coding. Cari benchmark untuk lihat yg terbaik. Yg populer benchmark pakai Human Eval dan MBPP

Human Eval = Dibuat OpenAI, it consists of 164 original programming problems, assessing language comprehension, algorithms, and simple mathematics, with some comparable to simple software interview questions.

MBPP = Dibuat Google, the benchmark consists of around 1,000 crowd-sourced Python programming problems, designed to be solvable by entry level programmers, covering programming fundamentals, standard library functionality, and so on.

Bisa lihat hasil benchmark di :

⚠️ Perhatikan kecepatan generate data (token/second), pricing dan accuracy dalam menentukan LLM

Lalu cari modelnya di Together AI, dia hosting LLM Deepseek Coder 33B di cloud, biaya kirim-terima text USD 0.8/1M token, 1M token sekitar 750.000 kata. Bisa dibilang API Open AI untuk LLM open sources

Prices are per 1 million tokens including input and output tokens for Chat, Language and Code models, https://www.together.ai/pricing