As of: 2026-11-19 22:11 CET. Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/ Canonical page: https://postcutoff.com/m/ling-3-1-flash/ # Ling 3.1 Flash inclusionAI (Ant Group), Ling 3 family. Status: current. Type: reasoning model. Released: 2026-09-29. Training cutoff: not published. ## How to call it | Provider | Model id | Endpoint | Docs | |---|---|---|---| | OpenRouter | `inclusionai/ling-3.1-flash` | | https://openrouter.ai/inclusionai/ling-3.1-flash | | Vercel AI Gateway | `inclusionai/ling-3.1-flash` | | https://vercel.com/ai-gateway/models/ling-3.1-flash | | Hugging Face | `inclusionAI/Ling-3.1-flash` | | https://huggingface.co/inclusionAI/Ling-3.1-flash | | Hugging Face (Oct 27 update) | `inclusionAI/Ling-3.1-flash-1019` | | https://huggingface.co/inclusionAI/Ling-3.1-flash-1019 | | Hugging Face (base model) | `inclusionAI/Ling-3.1-flash-base` | | https://huggingface.co/inclusionAI/Ling-3.1-flash-base | ## Pricing - Input: $0 per 1M tokens - Output: $0 per 1M tokens - Unit as published: per 1M tokens (USD); free trial on gateways (Vercel lists free through 2026-10-13) - Source: https://openrouter.ai/inclusionai/ling-3.1-flash - Last checked: 2026-10-28 ## Specs - Input: text - Output: text - Context window: 1,000,000 tokens - Max output: 32,768 tokens - Open weights: yes - Licence: mit ## What it has not seen - No published training cutoff. Counted from its release (2026-09-29): at least 1,385 AI events since (302 major, 58 historic), as of 2026-11-19. - Everything since August 2026: https://postcutoff.com/since/2026-08/ ## What stands out - 560B-total / 25B-active hybrid-reasoning MoE: Sparse MoE for coding, agents and tool use; reasoning enabled by default (can be turned off via the reasoning parameter on OpenRouter); tools and tool_choice supported. Source: https://openrouter.ai/inclusionai/ling-3.1-flash ## Notes Open weights released Oct 13–14, 2026 on Hugging Face (MIT, 1M context, hybrid-linear MoE: 63 KDA + 9 gated MLA layers; SGLang day-0 support on 8x H200 or 4x GB300). The Sept 29 trial was capped at 256K context; gateway trial pricing was free (Vercel through Oct 13); no official paid pricing yet. Served on Vercel by Novita. Oct 27: Ling-3.1-flash-1019 checkpoint (post-training update; model card reports AA Intelligence Index 44, up from 41) and Ling-3.1-flash-base. Free trial model as of Oct 2, 2026. Example (OpenRouter, OpenAI-compatible): ```bash curl https://openrouter.ai/api/v1/chat/completions -H "Authorization: Bearer $OPENROUTER_API_KEY" \ -H "Content-Type: application/json" -d '{"model":"inclusionai/ling-3.1-flash","messages":[{"role":"user","content":"Hello"}]}' ``` Self-hosting (model card, Oct 14, 2026): `docker pull lmsysorg/sglang:dev-ling-3.1-flash`, then `python3 -m sglang.launch_server --model-path inclusionAI/Ling-3.1-flash --tp 8 --ep 8 --speculative-algorithm NEXTN`. Recommended sampling: temperature 1.0, top_p 0.95, top_k 20, `enable_thinking: true`. ### Changelog - 2026-10-14: open weights (MIT, 1M context) released; status, license, access and notes updated - 2026-10-28: added Ling-3.1-flash-1019 and base checkpoints Model page: https://postcutoff.com/m/ling-3-1-flash/