Best LLMs for Infrastructure & Utility
Mechanical competence at format conversion, metadata manipulation, prompt rewriting, translation; minimal domain expertise required.
Mechanical competence at format conversion, metadata manipulation, prompt rewriting, translation; minimal domain expertise required.
Task-by-task breakdown
Research Query Generation
Generates platform-specific research search queries for a subject. Pooled across Google, Reddit, X/Twitter, Bluesky, Hacker News, OpenAlex, and generic web — each variant honours its platform's …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| GPT-5.6 Luna ★ | 93% | HIGH | best value |
| Qwen 3.5 Flash | 94% | MEDIUM | 1x |
| GPT-5.6 Terra | 96% | HIGH | 2.7x |
| Gemini 3.1 Pro Preview | 100% | HIGH | 3.2x |
| Qwen 3.6 Plus | 90% | MEDIUM | 4.3x |
| Kimi K2.6 | 99% | HIGH | 5.6x |
| GPT-5.6 Sol | 98% | HIGH | 5.7x |
| GPT-5.5 best | 100% | MEDIUM | 9.9x |
Research Query Validation
Validates and optimises a research search query against its target platform's syntax rules. Strips unsupported operators (boolean OR on Bluesky, parentheses, etc.) and removes year numbers, region …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| MiniMax M3 ★ best | 100% | RANKED | best value |
| GPT-5.6 Luna | 92% | MEDIUM | 1.8x |
| Gemini 3.5 Flash | 97% | HIGH | 5.5x |
| GPT-5.5 | 100% | HIGH | 18x |
Metadata Paragraph Rewriting
Polishes a draft metadata paragraph (research period, source counts, source types) at the head of a report. Preserves all facts and approximate length; only the wording is rewritten for clarity and …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| Claude Sonnet 5 ★ best | 100% | RANKED | best value |
| Gemini 3.5 Flash | 96% | HIGH | 3.5x |
Onboarding Chapter Prompt Adaptation
Adapts generic chapter outlines into subject-specific system and user prompts for the synthesis pipeline. Honours variable-substitution syntax for sources, subject_name, subject_code, and …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| DeepSeek V4 Flash ★ | 91% | HIGH | best value |
| GPT-5.4 Nano | 90% | HIGH | 2.3x |
| DeepSeek V4 Pro | 92% | MEDIUM | 5.5x |
| Qwen 3.6 Plus | 96% | RANKED | 12x |
| Kimi K2.6 | 99% | RANKED | 20x |
| Claude Sonnet 4.6 | 96% | HIGH | 41x |
| GPT-5.5 best | 100% | RANKED | 90x |
Report Image Generation
Generates a cover/social image from a report-derived image prompt. Wraps DALL-E / Imagen / Stable Diffusion / etc. behind a single task-type for batched dispatch.
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| Gemini 3.1 Flash Image Preview ★ | 92% | HIGH | best value |
| GPT-image-2 | 96% | RANKED | 1.5x |
| Gemini 3 Pro Image Preview best | 100% | MEDIUM | 1.6x |
Markdown Newline Repair
Re-inserts blank lines into a markdown document whose line breaks were stripped (headings, separators, and paragraphs collapsed onto one line). Must not modify any non-whitespace character — pure …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| MiniMax M3 ★ | 91% | MEDIUM | best value |
| Gemini 3.5 Flash best | 100% | RANKED | 32x |
Translation
Translates a batched list of text items into clear, natural English (or another target language) while preserving meaning, hashtags, @mentions, URLs, and inline tokens. Returns items in the same …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| DeepSeek V4 Flash ★ best | 100% | RANKED | best value |
| Gemini 3.5 Flash | 94% | MEDIUM | 21x |
| Claude Sonnet 5 | 90% | RANKED | 30x |
Prompt Adaptation
Renormalizes a pool/template prompt for a specific (capability x model). Never itself adapted (adaptation_exempt); recursion is also backstopped by the _ADAPTING re-entrancy guard. Outputs ARE …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| Tencent Hy3 ★ | 98% | RANKED | best value |
| DeepSeek V4 Flash | 91% | MEDIUM | 2.7x |
| Qwen 3.5 Flash | 91% | RANKED | 3.6x |
| GPT-5.6 Luna | 95% | RANKED | 4.3x |
| DeepSeek V4 Pro | 96% | RANKED | 4.5x |
| NVIDIA Nemotron-3 Ultra 550B | 93% | MEDIUM | 5.2x |
| GPT-5.6 Terra | 93% | HIGH | 8.2x |
| Claude Sonnet 4.6 best | 100% | RANKED | 13x |
| Gemini 3.5 Flash | 98% | RANKED | 16x |
| GPT-5.6 Sol | 95% | HIGH | 21x |
| Gemini 3.1 Pro Preview | 98% | RANKED | 24x |
| Qwen 3.6 Plus | 99% | RANKED | 26x |
| Meta Muse Spark 1.1 | 98% | HIGH | 27x |
| Kimi K2.6 | 98% | HIGH | 47x |
| GPT-5.5 | 93% | RANKED | 104x |
Claim Refinement
Reviews extracted claims and either REFINES (adds missing subject, expands abbreviations, resolves ambiguous references) or DROPS (methodology mentions, promotional content, claims about the source …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| Gemini 3.1 Flash Lite ★ | 100% | RANKED | best value |
| MiniMax M3 | 90% | MEDIUM | 2.5x |
| Tencent Hy3 best | 100% | HIGH | 3.2x |
| Qwen 3.5 Flash | 92% | HIGH | 5.1x |
| GPT-5.6 Luna | 93% | MEDIUM | 5.2x |
| Qwen 3.7 Plus | 98% | RANKED | 7.9x |
| Gemini 3.5 Flash | 98% | HIGH | 13x |
| Qwen 3.6 Flash | 96% | MEDIUM | 14x |
| Claude Sonnet 5 | 99% | HIGH | 15x |
| Qwen 3.6 Plus | 96% | HIGH | 16x |
| Claude Opus 4.8 | 94% | MEDIUM | 35x |
Image Prompt Generation
Transforms report content into an image-generation prompt for DALL-E / Imagen / similar. Specifies visual style, subject focus, colour palette, composition, mood, and technical details suitable for a …
| Model | Quality (% of best) | Confidence | Overpay |
|---|---|---|---|
| DeepSeek V4 Flash ★ | 93% | RANKED | best value |
| Tencent Hy3 | 91% | HIGH | 2.9x |
| MiniMax M3 | 95% | MEDIUM | 5x |
| Qwen 3.5 Flash | 92% | RANKED | 5.3x |
| GPT-5.6 Luna | 98% | RANKED | 6.5x |
| GPT-5.4 Mini | 91% | RANKED | 7.2x |
| DeepSeek V4 Pro | 94% | RANKED | 7.6x |
| NVIDIA Nemotron-3 Ultra 550B | 93% | HIGH | 11x |
| Qwen 3.7 Plus | 91% | RANKED | 13x |
| GPT-5.6 Terra best | 100% | HIGH | 14x |
| Qwen 3.6 Flash | 92% | RANKED | 16x |
| Qwen 3.6 Plus | 94% | RANKED | 18x |
| Grok 4.5 | 92% | RANKED | 19x |
| Claude Sonnet 5 | 94% | RANKED | 21x |
| Meta Muse Spark 1.1 | 97% | HIGH | 30x |
| Gemini 3.5 Flash | 96% | RANKED | 30x |
| Kimi K2.6 | 92% | RANKED | 33x |
| GPT-5.6 Sol | 99% | HIGH | 34x |
| Claude Sonnet 4.6 | 90% | RANKED | 51x |
| GPT-5.5 | 92% | RANKED | 90x |
Confidence — how sure we are about the quality score (more judgments + more agreement = higher confidence): RANKED many independent judges scored this model's outputs and their agreement is very high (most confident) — HIGH many judges have scored it and they mostly agree (well-pinned) — MEDIUM enough judges have weighed in to publish, but they disagree more than we'd like (treat with a small grain of salt). LOW-confidence cells are hidden everywhere on the site. See the methodology for the exact thresholds.