Model family and instruction-tuning approach appear better predictors of tutoring quality than parameter count alone¶
Claim · All claims
Evidence · 1 study · designr2·q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Within the 11-model sample, qwen3.5 at 9B parameters (79%) outperformed the 120B nemotron-3-super (73%), and deepseek-r1 at 8B (77%) scored well above the 30B qwen3-coder (52%), suggesting family-level differences influence tutoring quality better than size alone. → H. Chad Lane 2026
Evidence¶
H. Chad Lane 2026¶
H. Chad Lane, Bryson Kageler. (2026). CSTutorBench: Benchmarking Small Language Models as Tutors for Block-Based Programming. SLM4ED'26: The 1st Workshop of Small Language Models for Education. https://github.com/InviteInstitute/CSTutorBench doi:10.1145/3770761.3777333
q2 · i? · design · r2
Comparison of overall Trial 2 scores across 11 models from six families (Table 2). The top three were gemma-4:31b (89%), qwen3.5:9b (79%), and qwen3.6 (78%); the weakest included olmo-3:7b (49%) and gemma3:4b (50%). The authors state they cannot cleanly disentangle family, instruction tuning, and parameter count with only 11 models.
"Notably, qwen3.5 achieves this with only 9B parameters, outperforming the 120B nemotron-3-super (73%) by 6 percentage points."
Discussion¶
Related Claims¶
- CDPK performance scales with model size, with a sharp Pareto-frontier drop-off below around 8B parameters — related
- Under identical LoRA hyperparameters and training data, the 27B Gemma model outperforms the 70B LLaMA model on every computed essay-scoring metric — a narrower finding that bears on this claim
- A research-grounded prompt revision improved scores for 10 of 11 models, with gains of 6.6 to 16.2 percentage points — related
- Model scale does not consistently improve pedagogical safety: within the Qwen2.5 family the 72B model beats the 7B on only 17 of 33 subject-dimension pairs — a narrower finding that bears on this claim
- Adaptation gains grow monotonically with model capacity under the same LoRA recipe, though three data points cannot establish a scaling law — related
- Education-oriented instruction tuning consistently improves curriculum grounding, K-12 problem solving, and pedagogical tutoring at every evaluated model scale — related
- Six AI tutors individually matched human tutoring, with Gemma 4 31B equivalent at 918× lower cost per percentage point of learning gain — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
model-family-beats-parameter-count-for-tutoring-quality |
title |
Model family and instruction-tuning approach appear better predictors of tutoring quality than parameter count alone |
description |
Model family and instruction-tuning approach appear better predictors of tutoring quality than parameter count alone |
status |
review |
evidence_strength |
weak |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-10 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — h-chad-lane-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.