🏆 Quality winner: openrouter:minimax/minimax-m2.7
c2b157 (minimax) combines a fully complete, accurate rewritten passage, a rich six-term glossary, and a fair 10-question quiz whose vocab and evidence-pairing items are all answerable from its own passage, with analogies (world-class university, F1 car) that genuinely share the copy-the-output mechanism. Its closest rivals e2507b and 2eb734 are equally strong, but 2eb734's passage and e2507b are also excellent; c2b157 edges ahead on the balance of completeness (no truncation, unlike 0313fc, a95998, and haiku d7fde8-style truncations) and precise, safety-aware glossary. Several otherwise-strong runs (f95e70 one question, a0132a three questions, fa777b generic analogies) fall behind on quiz depth or analogy fidelity.
💰 Best value: openrouter:deepseek/deepseek-v4-pro
— quality 9/10 at 0.62¢ → value 8.7/10
Value = 0.7·quality + 0.3·cheapness. Quality stays dominant, so a cheap-but-weak output can't win on price alone.
| # | Model | Facts | Analogy | Age-fit | Clarity | Quiz | Glossary | Passage | Overall | Cost | Value | Notes |
|---|
| 🥇 |
openrouter:deepseek/deepseek-v4-pro
view ↗ |
10 | 9 | 9 | 9 | 9 | 9 | 9 |
9 |
0.62¢ |
8.7 |
Excellent fact grounding with the Tu-4/B-29 Cold War analogy sharing the same reverse-engineering mechanism. Full 10-question quiz with strong SAT-style vocab and evidence-pairing items. Rewritten passage is accurate and engaging. |
| 🥈 |
openrouter:minimax/minimax-m2.7
view ↗ |
10 | 8 | 9 | 9 | 9 | 9 | 9 |
9 |
0.62¢ |
8.7 |
Very accurate, well-structured, with a good glossary including safety guardrails. Quiz is fair and answerable from the passage; vocab items ('leverage','surreptitious') tie back to text. Analogies (university, F1 car) share the copying mechanism. |
| 🥉 |
openrouter:moonshotai/kimi-k2.6
view ↗ |
9 | 9 | 9 | 9 | 9 | 9 | 9 |
9 |
9.62¢ |
6.6 |
Conservatory/pharma analogies share the incentive-structure mechanism well. Quiz vocab ('frontier','lightweight') and evidence pairing solid. Rewritten passage is complete, accurate, and elegantly written—one of the strongest. |
| 4 |
anthropic:claude-haiku-4-5-20251001
view ↗ |
9 | 8 | 9 | 9 | 9 | 9 | 8 |
9 |
2.57¢ |
7.5 |
Nuclear arms race and doping analogies fit; explains the 'legal vs illegal' distinction cleanly. Strong 10-question quiz with good vocab and evidence pairing. Rich glossary. Rewritten passage accurate but truncated at the very end. |
| 5 |
openrouter:mistralai/mistral-large-2512
view ↗ |
9 | 8 | 9 | 9 | 8 | 8 | 8 |
8 |
0.85¢ |
8.0 |
Solid and faithful; 'save scumming' and supply-chain heist analogies are decent. Q5 uses 'surreptitious' as a vocab word but the term doesn't appear in the rewritten passage, a minor answerability issue. Strong prose. |
| 6 |
openrouter:stepfun/step-3.7-flash
view ↗ |
9 | 9 | 9 | 8 | 8 | 8 | 7 |
8 |
2.09¢ |
6.8 |
Strong analogies (SAT tutor, burger recipe) with matching mechanism. Full quiz with trap taxonomy. However the rewritten_passage is truncated mid-sentence, hurting reading_passage_quality. |
| 7 |
openrouter:google/gemini-3.1-flash-lite
view ↗ |
9 | 8 | 8 | 9 | 8 | 8 | 8 |
8 |
0.59¢ |
8.0 |
Chef recipe and capture-the-flag analogies fit reasonably. Full 10-question quiz, clean structure. Accurate rewritten passage. Solid all-around but glossary is 4 terms and slightly generic. |
| 8 |
openrouter:x-ai/grok-4.3
view ↗ |
9 | 7 | 9 | 9 | 8 | 8 | 9 |
8 |
1.64¢ |
7.4 |
Clean, accurate, complete passage. Playbook-scouting analogy is fine but student-copying is generic. Full quiz answerable from the passage; vocab items grounded. Reliable but analogies less distinctive than top runs. |
| 9 |
openrouter:openai/gpt-5.4-nano
view ↗ |
9 | 8 | 8 | 8 | 8 | 8 | 8 |
8 |
0.63¢ |
8.0 |
Careful, hedged ('the article says') framing keeps facts tight. Textbook/scouting analogies fit. Quiz well-constructed with grounded vocab. Slightly repetitive meta-referencing ('the article') in the passage but accurate. |
| 10 |
openrouter:z-ai/glm-5-turbo
view ↗ |
8 | 8 | 9 | 9 | 6 | 9 | 7 |
7 |
3.40¢ |
6.1 |
Music-producer and scouting analogies are strong. However Q4 tests the word 'sharp' via a phrase ('drew a sharp distinction') that does not appear in the truncated passage—an answerability failure. Rewritten passage is also truncated. |
| 11 |
openrouter:bytedance-seed/seed-2.0-lite
view ↗ |
8 | 7 | 7 | 8 | 7 | 8 | 7 |
7 |
1.73¢ |
6.7 |
MIT-lab and Twitch-scraping analogies are decent. Q4 tests 'principal' but the rewritten passage uses 'principally' rather than 'principal' as adjective—minor mismatch. Otherwise complete and reasonable. |
| 12 |
openrouter:qwen/qwen3-max-thinking
view ↗ |
8 | 6 | 8 | 7 | 5 | 8 | 9 |
6 |
1.02¢ |
6.0 |
Good rewritten passage and glossary, but only ONE quiz question provided, severely undercutting quiz_quality. Analogies (answer key, handbag counterfeit) are okay but the handbag one shares less mechanism. |
| 13 |
openrouter:amazon/nova-pro-v1
view ↗ |
8 | 5 | 6 | 6 | 6 | 6 | 6 |
6 |
1.34¢ |
6.0 |
Faithful but thin. Chess and doping analogies are generic and don't capture the distillation mechanism. Short glossary (3 terms), brief passage. Q10 quotes a caption-style line as evidence somewhat weakly. |
| 14 |
openrouter:meta-llama/llama-4-maverick
view ↗ |
8 | 5 | 7 | 7 | 7 | 7 | 7 |
6 |
0.26¢ |
7.2 |
Faithful but generic analogies (copying test answers, stealing lesson plans). Quiz functional but somewhat shallow. Serviceable rewritten passage. Nothing stands out. |
| 15 |
openrouter:baidu/ernie-4.5-vl-424b-a47b
view ↗ |
8 | 7 | 7 | 7 | 5 | 6 | 7 |
6 |
0.43¢ |
7.2 |
Chef-recipe and counterfeit analogies fine, but only 3 quiz questions and a 3-term glossary make it feel abbreviated. Q2 defines 'distil' as 'replicate' over 'steal' somewhat loosely. Passage accurate but compressed. |
Scores 0–10 · 🟩 ≥8 · 🟨 4–7 · 🟥 <4 · sorted by Overall. Hover a column header for its full rubric name.
| # | Model | Quality | Cost | Value |
|---|
| 🥇 |
openrouter:deepseek/deepseek-v4-pro |
9 |
0.62¢ |
8.7 |
| 🥈 |
openrouter:minimax/minimax-m2.7 |
9 |
0.62¢ |
8.7 |
| 🥉 |
openrouter:mistralai/mistral-large-2512 |
8 |
0.85¢ |
8.0 |
| 4 |
openrouter:google/gemini-3.1-flash-lite |
8 |
0.59¢ |
8.0 |
| 5 |
openrouter:openai/gpt-5.4-nano |
8 |
0.63¢ |
8.0 |
| 6 |
anthropic:claude-haiku-4-5-20251001 |
9 |
2.57¢ |
7.5 |
| 7 |
openrouter:x-ai/grok-4.3 |
8 |
1.64¢ |
7.4 |
| 8 |
openrouter:meta-llama/llama-4-maverick |
6 |
0.26¢ |
7.2 |
| 9 |
openrouter:baidu/ernie-4.5-vl-424b-a47b |
6 |
0.43¢ |
7.2 |
| 10 |
openrouter:stepfun/step-3.7-flash |
8 |
2.09¢ |
6.8 |
| 11 |
openrouter:bytedance-seed/seed-2.0-lite |
7 |
1.73¢ |
6.7 |
| 12 |
openrouter:moonshotai/kimi-k2.6 |
9 |
9.62¢ |
6.6 |
| 13 |
openrouter:z-ai/glm-5-turbo |
7 |
3.40¢ |
6.1 |
| 14 |
openrouter:qwen/qwen3-max-thinking |
6 |
1.02¢ |
6.0 |
| 15 |
openrouter:amazon/nova-pro-v1 |
6 |
1.34¢ |
6.0 |
| # | Model | Quality | Cost | Value |
|---|
| 1 |
openrouter:deepseek/deepseek-v4-pro |
9 |
0.62¢ |
8.7 |
| 2 |
openrouter:minimax/minimax-m2.7 |
9 |
0.62¢ |
8.7 |
| 3 |
anthropic:claude-haiku-4-5-20251001 |
9 |
2.57¢ |
7.5 |
| 4 |
openrouter:mistralai/mistral-large-2512 |
8 |
0.85¢ |
8.0 |
| 5 |
openrouter:google/gemini-3.1-flash-lite |
8 |
0.59¢ |
8.0 |
| 6 |
openrouter:openai/gpt-5.4-nano |
8 |
0.63¢ |
8.0 |
| 7 |
openrouter:x-ai/grok-4.3 |
8 |
1.64¢ |
7.4 |
| 8 |
openrouter:stepfun/step-3.7-flash |
8 |
2.09¢ |
6.8 |
| 9 |
openrouter:bytedance-seed/seed-2.0-lite |
7 |
1.73¢ |
6.7 |
| 10 |
openrouter:meta-llama/llama-4-maverick |
6 |
0.26¢ |
7.2 |
| 11 |
openrouter:baidu/ernie-4.5-vl-424b-a47b |
6 |
0.43¢ |
7.2 |
| 12 |
openrouter:qwen/qwen3-max-thinking |
6 |
1.02¢ |
6.0 |
| 13 |
openrouter:amazon/nova-pro-v1 |
6 |
1.34¢ |
6.0 |
Strong quality without paying flagship prices — the cheap-and-good picks.