Valiant Labs

At Valiant Labs, we build frontier intelligence and release it open source. For everyone to use.

Esper 3.1 is finetuned on high-difficulty coding, architecture, DevOps, and MLOps:

ValiantLabs/Qwen3-4B-Thinking-2507-Esper3.1

ValiantLabs/gpt-oss-20b-Esper3.1

Shining Valiant 3 is finetuned on high-difficulty science-reasoning, LLMOps and AI architecture, and general reasoning:

ValiantLabs/Qwen3-1.7B-ShiningValiant3

ValiantLabs/Qwen3-4B-ShiningValiant3

ValiantLabs/Qwen3-8B-ShiningValiant3

ValiantLabs/gpt-oss-20b-ShiningValiant3

Esper 3 is finetuned on high-difficulty coding, architecture, DevOps, and general reasoning:

ValiantLabs/Qwen3-4B-Esper3

ValiantLabs/Qwen3-8B-Esper3

ValiantLabs/Qwen3-14B-Esper3

ValiantLabs/DeepSeek-R1-0528-Qwen3-8B-Esper3

Cobalt 2 is finetuned on high-difficulty math and general reasoning:

ValiantLabs/Qwen3-14B-Cobalt2

Shining Valiant 2 is finetuned on high-quality science, complex queries, and general chat:

ValiantLabs/Llama3.1-8B-ShiningValiant2

ValiantLabs/Llama3.1-70B-ShiningValiant2

ValiantLabs/Llama3.2-3B-ShiningValiant2

Check out our website!

Contact us via email at [email protected]