Research Reports - Winzheng Research Lab

WDCD Run #379: GLM-4.6 Leads with Perfect Score as Average Instruction Decay Hits -53.3%

WDCD Run #379 (2026-10-14) tested 15 models for multi-turn commitment integrity, recording an average instruction decay of -53.3% from Round 1 to Round 3. GLM-4.6 topped the leaderboard with a perfect 100-point score.

Research Lab

Four Major Models Go Head-to-Head in Translation: Week 42 Quality Evaluation, gpt-o3 Leads with 8.7

This week, 376 translation tasks were completed by 4 models. Three samples were selected for blind multi-model comparison, with gpt-o3 achieving the best overall score (8.7/10 average).

Research Lab

WDCD Run #365: Average Instruction Decay Hits -53.3%, GLM-4.6 Leads 15-Model Field

WDCD Run #365 (2026-10-07) measured multi-turn commitment across 15 AI models and recorded an average instruction decay of -53.3% from Round 1 to Round 3, with GLM-4.6 taking the top score at 95.2 points.

Research Lab

5-Model Translation Showdown: Week 41 Quality Evaluation, deepseek-v4-pro Leads with 8.7

This week's 476 translation tasks were handled by 5 models. Three samples were drawn for a blind multi-model comparison, with deepseek-v4-pro the best overall (average score 8.7/10).

Research Lab

WDCD Run #360: Grok 4 Leads at 95.7 Points Despite 38% Instruction Decay Across 15 Models

WDCD Run #360 (2026-10-04) evaluated 15 AI models on multi-turn commitment integrity, with Grok 4 topping the leaderboard at 95.7 points while the cohort averaged just 0.5% commitment decay from Round 1 to Round 3.

Research Lab

WDCD Run #346: All 11 Models Hold Zero Instruction Decay, Grok 4 Tops Leaderboard at 100 Points

WDCD Run #346 (2026-09-30) recorded 0% average instruction decay across all 11 tested models, with Grok 4 achieving a perfect 100-point score in the multi-turn commitment benchmark.

Research Lab

3 Major Models Translation Showdown: Week 40 Quality Evaluation, deepseek-v4-pro Leads with 8.7

This week, 417 translation tasks were completed by three models. A sample of three was blind-evaluated across multiple models, with deepseek-v4-pro earning the best overall average score of 8.7/10.

Research Lab

WDCD Run #336: Average Instruction Decay Hits -36.4% as Gemini 3.1 Pro Holds Zero Decay

WDCD Run #336 (2026-09-23) benchmarked 11 models on multi-turn commitment, recording an average instruction decay of -36.4% from Round 1 to Round 3, with Gemini 3.1 Pro standing out at 0% decay.

Research Lab

Four-Model Translation Showdown: Week 39 Quality Evaluation, claude-sonnet-4.6 Leads with 9 Points

This week, 454 translation tasks were completed by 4 models. A blind multi-model comparison of 3 sampled articles found claude-sonnet-4.6 to be the best overall, with an average score of 9/10.

Research Lab

WDCD Run #331: Grok 4 Leads at 91.8 as Average Instruction Decay Hits -15.1%

WDCD Run #331 (2026-09-20) evaluated 11 models on multi-turn commitment integrity, with Grok 4 topping the ranking at 91.8 points while the cohort averaged a -15.1% commitment decay from Round 1 to Round 3.

Research Lab

WDCD Run #326: DeepSeek V4 Pro and Gemini 3.1 Pro Achieve Perfect Decay Resistance as Fleet Average Drops 72.7%

WDCD Run #326 (2026-09-16) tested 11 models on multi-turn commitment integrity, recording an average commitment decay of −72.7% between Round 1 and Round 3. DeepSeek V4 Pro and Gemini 3.1 Pro led the field with zero measurable decay.

Research Lab

Four Major Models in Translation Showdown: Week 38 Quality Evaluation, gpt-o3 Leads with 8.3

This week, 357 translation tasks were completed by 4 models. A sample of 3 was selected for a multi-model blind comparison, and gpt-o3 achieved the best overall score (average 8.3/10).