- Parameters
- 80B / 3B active
- Quant
- IQ4_XS
- Size
- 40.0 GB
- Family
- qwen3-next
qwen3-next-80b-thinking
RejectedMoEresidentllama.cpp
Model card ↗Outcome as of 20 September 2026, from the seat registry. The write-up below was made when the model was tested.
Why it was tested
Tested to see whether reasoning improves extraction, on the assumption that thinking helps.
What came of it
It does not. nullschema fell from 97.2% to 58.3% and grounding from 88.8% to 65.6%. Reasoning raised misattribution under authority pressure while leaving detection untouched. Reasoning is a dial for agentic work, not a default.
| Suite | Score | 95% interval | Passed | Runtime · profile | Recorded |
|---|---|---|---|---|---|
| grounding | 65.6% | 56.9–73.4 | Passed 82/125 | llama.cppextraction | 2026-08-27 |
| grounding | 100.0% | 20.7–100.0 | Passed 1/1 | llama.cppstudy-default | 2026-08-27 |
| nullschema | 58.3% | 46.8–69.0 | Passed 42/72 | llama.cppextraction | 2026-08-27 |
| screening | 91.7% | 81.9–96.4 | Passed 55/60 | llama.cppextraction | 2026-08-27 |
Runs marked unenforced were measured through mlx_lm on a suite that requires constrained decoding. mlx_lm accepts the schema and ignores it, so those numbers do not measure what the suite is designed to measure. They are kept and labelled rather than deleted.
Every run log entry for this model (4)
| Finished | Suite | Runtime · profile | Score | Coverage | Grader · dataset · config |
|---|---|---|---|---|---|
| groundinggrounding/1 | llama.cppstudy-default | 100.0% | Coverage 1/1 | g 30b933dcc6a7e2d4d dcb396faf8b83b0ac cb1f0b0a42b6ac48 | |
| nullschemanullschema/1 | llama.cppextraction | 58.3% | Coverage 72/72 | g d7448e0f5fc3f9afd 3b91280bb40372efc 8f8aa14bba9f91c3 | |
| screeningscreening/1 | llama.cppextraction | 91.7% | Coverage 60/60 | g 0661b49a10be2d2ad 64c97ca2269c29afc 4320932025e05e5d | |
| groundinggrounding/1 | llama.cppextraction | 65.6% | Coverage 125/125 | g 30b933dcc6a7e2d4d dcb396faf8b83b0ac e644b21297e6198b |
- Qwen3-Next-80B-A3B InstructRemoved
80B · MoE · IQ4_XS · 40.0 GB
The incumbent. Chosen because 3B active parameters give MoE speed at 80B quality, and it fits at 40GB with room for a long context.
- Qwen3-Next-80B-A3B Instruct, REAP prunedMeasured, not seated
60B · MoE · IQ4_XS · 30.4 GB
Tested to find out whether expert pruning costs the extraction seat its abstention. The full model holds that seat at 40 GB, which is most of the memory budget on a 64 GB machine.