Ling-3.1-flash has public MIT weights and SGLang/vLLM deployment recipes.
InclusionAI releases Ling-3.1-flash weights for long-horizon agent work
2026-10-13Frontier watchWATCH
InclusionAI released MIT-licensed Ling-3.1-flash weights. The model has 560B total parameters with 25B active per token and targets coding, search, office and general agent work. Official SGLang and vLLM recipes are public, but representative deployment still requires eight H200s or four GB300s.
What this means for you
For teams evaluating open-weight coding agents, this is a deployable candidate rather than an API-only leaderboard entry. The 1M context remains a design claim: the repository config says 262,144 and the LingCode harness is not public, so long-context and coding results need reproduction.
FRONTIER WATCH
Public artifacts and validated results are different.
The recorded fact is the public work and its artifacts. Author-reported outcomes do not establish independent reproduction. Next checks include usable code, task conditions and third-party validation.
Evidence & related material
Technical details & scores
| Architecture | hybrid-linear MoE; 560B total, 25B active |
|---|---|
| License | MIT; public FP8 safetensors, about 574 GB |
| Serving | official SGLang and vLLM recipes; typical path uses 8x H200 or 4x GB300 |
| Limits | model card claims 1M design, repository config is 262144; author benchmarks; LingCode harness not public |
| Time Basis | HF repository created 2026-10-13T11:16:41Z; major weight uploads followed at 15:35Z |
Technical impact 4
Engineering value 4
Adoption signal 2
Maturity 2
Verification cost 4
Risk 3