Models

Super AI Engineer LLM model family

Model variants, evaluation context, and deployment-ready engineering information.

Model overview

The Super AI Engineer LLM family is built for Thai-first engineering workflows with curated Thai + multilingual data. It ships in three variants: base, instruct, and chat, covering everything from continued pretraining to production chat.

License

Released under Apache-2.0, free for research and commercial use. Please review the model card for intended use and limitations before deploying.

Available versions

Thai LLM Base v0.1

thai-llm-base-v0.1

active
Type
base
Parameters
8B
Context
8,192 tokens
Max output
4,096
License
Apache-2.0
Backends
B200 LANTA

Thai LLM Instruct v0.1

thai-llm-instruct-v0.1

active
Type
instruct
Parameters
8B
Context
8,192 tokens
Max output
4,096
License
Apache-2.0
Backends
B200 LANTA

Thai LLM Chat v0.1

thai-llm-chat-v0.1

active
Type
chat
Parameters
8B
Context
16,384 tokens
Max output
4,096
License
Apache-2.0
Backends
B200 LANTA

taPan Model (Qwen 1B pruned TH)

taPanModel

active
Type
chat
Parameters
1B
Context
4,096 tokens
Max output
2,048
License
Apache-2.0
Backends

SPAISS6 1M

spaiss6-1M

active
Type
chat
Parameters
1M
Context
512 tokens
Max output
256
License
Apache-2.0
Backends