# Is Alpie Seed SOTA?

Pending benchmarks. Announced Nov 6, 2026; no independent results yet.

As of Nov 22, 2026, Alpie Seed is pending benchmarks on isitsota: announced Nov 6, 2026, with no independent results on the boards we track yet.

169Pi's 9B open-weight multimodal model, post-trained from Qwen3.5-9B with supervised fine-tuning and GRPO. It accepts text, images and video, with thinking and non-thinking modes and a 262,144-token native context. Apache-2.0 weights and GGUF builds support local/offline deployment; Ollama tags differ in input support. No separate public hosted API is confirmed.

Canonical: https://isitsota.com/m/alpie-seed
Last updated: 2026-11-07
Tracking since: 2026-11-07

## Announced facts

| Fact | Value |
| --- | --- |
| Lab | 169Pi |
| Announced | Nov 6, 2026 |
| API | Not yet |
| Open weights | Published |
| Parameters | 9B |
| Base | Qwen3.5-9B architecture retained; SFT and GRPO post-training rather than a structural redesign |
| Inputs | Text, image and video; thinking and non-thinking modes |
| Context | 262,144 tokens native context per model card; Ollama lists 256K |
| License | Apache-2.0 |
| Deployment | Public Hugging Face FP16 weights and 2/3/4-bit GGUF builds; official Ollama distribution and offline mobile demo. Ollama text tags and a separate vision tag have different input support |
| Serving | Self-hosting instructions for Transformers and vLLM; official post links a 169Pi Playground, but no separate public hosted API confirmed |
| Related | Separate from its Qwen3.5-9B base; quantized formats remain variants of Alpie Seed |

Tags marked unsure are not stated on the lab's own pages.

## What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

- [x] Publicly announced with a source
- [x] Usable by the public (API or weights)
- [ ] Listed on an independent board we track
- [ ] Enough results for a rank in a category
- [ ] Holds across two snapshots (crowns only)

## What the lab claims

Self-reported. Shown for context; not used in any verdict.

- Lab claim: FP16 checkpoint scores 82.54% MMLU-Pro and 84.32% GPQA Diamond with thinking enabled, max_tokens 65536 and temperature 0.0. These lab-reported scores do not apply to quantized variants; external comparisons were not rerun under matched settings.

## Sources

- LAB: [169Pi introduces Alpie Seed](https://x.com/169Pi_ai/status/2118663302679277891) (2026-11-06)
- LAB: [169Pi: Alpie Seed now available](https://x.com/169Pi_ai/status/2118663319326257577) (2026-11-06)
- REPO: [169Pi Alpie Seed model card](https://huggingface.co/169Pi/Alpie_seed)
- REPO: [169Pi Alpie Seed GGUF variants](https://huggingface.co/169Pi/Alpie_Seed_GGUF)
- LAB: [169Pi official Ollama distribution](https://ollama.com/169pi/alpie-seed)

Tracking continues until an independent board lists it.

## Related

- Lab: [169Pi](https://isitsota.com/labs/169pi.md)
- All pending releases: https://isitsota.com/releases.md
- JSON: https://api.isitsota.com/v1/models/alpie-seed
