MaLLaM π
Multi-lingual Malaysian Chat Language Model, 32k context length, Malaysian centric and private.
Multi-lingual
Able to understand Standard Malay, local Malay, Jawi, Standard English, Manglish, Mandarin and Indonesian as input.
Natively RAG
Integrate MaLLaM π with existing knowledge base for better User-Chat experiences, we use GARIS-PANDUAN-SRF-RFP-2024-v1.pdf for an example.
Function Call
Convert Natural text to JSON format for faster system integration, MaLLaM π achieved 100% JSON-able on NousResearch/json-mode-eval, higher is better.
Better Accuracy
Better accuracy compared to ChatGPT3.5 and open-source Malaysian Chat Models. We benchmarked on Malay test set, mesolitica/malay-llm-leaderboard, higher is better.
Playground
You can play around with MaLLaM π integrated with AI Safety, try it at Nous App
Try the API
Or if you want to integrate MaLLaM π with existing system and compatible with OpenAI library, Nous LLM Router Documentation
Pricing
Prepaid based, natively Multi-lingual, general knowledge, RAG, Function Call, Coding and Multi-turn.
Model name | Input / 1M tokens | Output / 1M tokens |
---|---|---|
MaLLaM π Tiny | MYR 2.50 | MYR 10.00 |
MaLLaM π Small | MYR 5.00 | MYR 15.00 |
Private
Self-host MaLLaM π in your private network for 100% privacy, either on-premise or private cloud, read more at MaLLaM π Self-hosted Enterprise
Frequently asked questions
What is the different this MaLLaM π and MaLLaM π open sourced?
This MaLLaM π continue pretraining on more dataset, finetuned with bigger instruction dataset and aligned with human policy.
What is the rate limit?
Currently we hard limit 100k Tokens per Minute.
How to topup?
Just go to billing page and topup! Minimum RM3 and Maximum RM99.