Purpose-trained single-pass generation completes a slide in a median of 17 seconds and an interactive page in 59 seconds over 220k production requests, and delivers artifacts at 15–22× lower per-artifact API cost than flagship models¶

Claim · All claims
Evidence · 1 study (2 entries) · design r3 · 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¶

q2 i? Over a seven-day live-traffic window the production models completed 180k slide generations at a median of 17s and 40k interactive pages at 59s. → CogEvol Team 2026 q2 i? At public list prices CogEvol-27B delivers near-flagship quality at 15–22× lower per-artifact API cost than Claude Opus 4.8 or GPT-5.4, and CogEvol-4B at ~100× lower. → CogEvol Team 2026 (2)

Evidence¶

CogEvol Team 2026¶

CogEvol Team, CogEvol Inc. & Tsinghua University. (2026). CogEvol: Towards Efficient and Reliable Learning Environment Generation. arXiv:2608.30968v2. https://arxiv.org/abs/2608.30968 doi:10.48550/arxiv.2608.30968

q2 · i? · design · r3

Production serving-database statistics (successful calls, seven-day window, August 2026) for the models serving OpenMAIC traffic. The report states these "are production numbers, not laboratory ones", with slides emitting ~1.8k output tokens and pages ~9k.

"the production models behind these medians—same architecture and parameter count as CogEvol-27B, and therefore the same serving speed—completed 180k slide generations at a median (P95) of 17s (26s) and 40k interactive pages at 59s (107s)."

CogEvol Team 2026 (2)¶

CogEvol Team, CogEvol Inc. & Tsinghua University. (2026). CogEvol: Towards Efficient and Reliable Learning Environment Generation. arXiv:2608.30968v2. https://arxiv.org/abs/2608.30968

q2 · i? · design · r3

Cost comparison computed from token usage on the identical benchmark runs at public list prices (Figure 1b); the report frames cost as deciding "who gets to use the technology" for under-resourced users.

"at public list prices, CogEvol-27B delivers near-flagship quality at 15–22× lower per-artifact API cost than Claude Opus 4.8 or GPT-5.4, and CogEvol-4B at∼100× lower."

Discussion¶

Related Claims¶

• On the report's internal suites CogEvol-27B scores 83.7 on slides and 63.7 on HTML-500 with zero hard failures, while no external flagship leads both modalities — related
• A deployment-hardening rehearsal with 50 distinct learner accounts recorded 50 completed sessions, 50 successful board pulls and 50 denials of student analytics access, with a median API session time of 3.064 s — related
• Larger dense generators passed smoke probes but all exceeded the +30% classroom latency limit in bake-off round 1 — related
• May 2026 token telemetry was cache-dominant: 82.9% of 73,950,305 recorded tokens were cache reads, suggesting persistent workflows shift the economic unit from cost per token to cost per completed artifact
• The OCR-based text input uses far fewer tokens than image-based multimodal input for the same slide, making it more cost-effective — related