Cost mode:

Category: Relevance, Classification & Matching · Rail: absolute · Typical I/O: 1980→3091 tokens

Models

Frontier on this task: DeepSeek V4 Flash at 8.78 / 10. Quality bar at 90%: 7.90.

point-estimate floor (CI low) · upper CI (less certain) · Bars sorted by blended cost; best-value model first. Greyed rows are MEDIUM+ models whose point estimate clears the bar but whose CI low does not.

ModelQuality scoreCI lowCost / 1k runsvs best value
Tencent Hy38.09 / 107.91$0.22best value
GPT-5.6 Luna8.46 / 108.32$0.231.1x more expensive
Gemini 3.5 Flash Lite8.60 / 108.35$0.301.4x more expensive
NVIDIA Nemotron-3 Nano 30B-A3B8.35 / 107.94$0.321.5x more expensive
Gemini 3.1 Flash Lite8.50 / 108.25$0.331.5x more expensive
MiniMax M38.40 / 108.33$0.873.9x more expensive
DeepSeek V4 Flash8.78 / 108.65$1.396.2x more expensive
Claude Haiku 4.58.06 / 107.67$1.577x more expensive
GLM-5.3 Flash8.70 / 108.53$1.788x more expensive
NVIDIA Nemotron 3.5 Lightning8.08 / 107.75$1.948.7x more expensive
GPT-5.6 Terra8.57 / 108.43$2.139.6x more expensive
NVIDIA Nemotron-3 Ultra 550B8.33 / 107.98$2.159.6x more expensive
NVIDIA Nemotron-3 Super 120B8.10 / 107.64$3.2314x more expensive
DeepSeek V4 Pro8.47 / 108.22$3.2314x more expensive
Qwen 3.8 Flash8.31 / 108.05$3.3715x more expensive
Gemini 3.8 Flash8.72 / 108.49$3.6316x more expensive
GPT-5.6 Sol8.48 / 108.34$3.9318x more expensive
Claude Sonnet 58.33 / 108.21$4.0918x more expensive
Thinking Machines Inkling Small8.74 / 108.52$4.5821x more expensive
Qwen 3.7 Plus8.33 / 108.16$5.1723x more expensive
Gemini 3.5 Flash8.49 / 108.39$7.7935x more expensive
Claude Opus 58.64 / 108.37$10.7148x more expensive
Meta Muse Spark 1.38.73 / 108.56$12.8558x more expensive
GLM-5.38.64 / 108.42$14.2164x more expensive
Thinking Machines Inkling8.58 / 108.31$19.7288x more expensive
Qwen 3.8 Max8.17 / 107.94$30.07135x more expensive
Tencent Hy4 Preview8.66 / 108.46$30.76138x more expensive
Grok 4.68.66 / 108.42$34.90157x more expensive
Moonshot Kimi K38.66 / 108.42$39.14176x more expensive
GPT-5.4 Nano7.71 / 107.36$0.231x more expensive

Cost breakdown

ModelQualityConfidenceCost / 1k runsOverpayMode
Tencent Hy3 OpenRouter8.09 / 10 CI [7.91, 8.27]RANKED$0.22best valuebatch
GPT-5.6 Luna OpenAI8.46 / 10 CI [8.32, 8.61]RANKED$0.231.1xbatch
Gemini 3.5 Flash Lite Gemini8.60 / 10 CI [8.35, 8.86]HIGH$0.301.4xbatch
NVIDIA Nemotron-3 Nano 30B-A3B OpenRouter8.35 / 10 CI [7.94, 8.77]MEDIUM$0.321.5xbatch
Gemini 3.1 Flash Lite Gemini8.50 / 10 CI [8.25, 8.75]HIGH$0.331.5xbatch
MiniMax M3 OpenRouter8.40 / 10 CI [8.33, 8.48]RANKED$0.873.9xbatch
DeepSeek V4 Flash best DeepSeek8.78 / 10 CI [8.65, 8.90]RANKED$1.396.2xbatch
Claude Haiku 4.5 Anthropic8.06 / 10 CI [7.67, 8.45]MEDIUM$1.577xbatch
GLM-5.3 Flash Z.AI8.70 / 10 CI [8.53, 8.87]RANKED$1.788xbatch
NVIDIA Nemotron 3.5 Lightning OpenRouter8.08 / 10 CI [7.75, 8.41]MEDIUM$1.948.7xbatch
GPT-5.6 Terra OpenAI8.57 / 10 CI [8.43, 8.71]RANKED$2.139.6xbatch
NVIDIA Nemotron-3 Ultra 550B OpenRouter8.33 / 10 CI [7.98, 8.68]MEDIUM$2.159.6xbatch
NVIDIA Nemotron-3 Super 120B OpenRouter8.10 / 10 CI [7.64, 8.57]MEDIUM$3.2314xbatch
DeepSeek V4 Pro DeepSeek8.47 / 10 CI [8.22, 8.72]HIGH$3.2314xbatch
Qwen 3.8 Flash Alibaba Cloud (DashScope)8.31 / 10 CI [8.05, 8.57]HIGH$3.3715xbatch
Gemini 3.8 Flash Gemini8.72 / 10 CI [8.49, 8.95]HIGH$3.6316xbatch
GPT-5.6 Sol OpenAI8.48 / 10 CI [8.34, 8.63]RANKED$3.9318xbatch
Claude Sonnet 5 Anthropic8.33 / 10 CI [8.21, 8.46]RANKED$4.0918xbatch
Thinking Machines Inkling Small OpenRouter8.74 / 10 CI [8.52, 8.97]HIGH$4.5821xbatch
Qwen 3.7 Plus Alibaba Cloud (DashScope)8.33 / 10 CI [8.16, 8.49]RANKED$5.1723xbatch
Gemini 3.5 Flash Gemini8.49 / 10 CI [8.39, 8.59]RANKED$7.7935xbatch
Claude Opus 5 Anthropic8.64 / 10 CI [8.37, 8.91]HIGH$10.7148xbatch
Meta Muse Spark 1.3 OpenRouter8.73 / 10 CI [8.56, 8.91]RANKED$12.8558xbatch
GLM-5.3 Z.AI8.64 / 10 CI [8.42, 8.86]HIGH$14.2164xbatch
Thinking Machines Inkling OpenRouter8.58 / 10 CI [8.31, 8.85]HIGH$19.7288xbatch
Qwen 3.8 Max Alibaba Cloud (DashScope)8.17 / 10 CI [7.94, 8.41]HIGH$30.07135xbatch
Tencent Hy4 Preview OpenRouter8.66 / 10 CI [8.46, 8.85]RANKED$30.76138xbatch
Grok 4.6 xAI8.66 / 10 CI [8.42, 8.90]HIGH$34.90157xbatch
Moonshot Kimi K3 Moonshot AI8.66 / 10 CI [8.42, 8.90]HIGH$39.14176xbatch

Overpay shows how much more you pay than the best-value model that clears the quality bar (marked ★) — the best-value good-enough option. "16x" means you overpay 16× — 16× that reference for no quality benefit above the bar. Typical call shape for this task: 1980 input tokens → 3091 output tokens, EMA-tracked from production traffic. Cost is the observed, all-in $ per 1,000 task runs: each model's own measured usage on this task — output verbosity, thinking/reasoning tokens, cache reads and writes, and the spend on its billed failures — priced at current list rates and adjusted by the billing overhead we actually reconcile against provider invoices. Models that answer tersely cost what they actually cost; models that think at length pay for it. Not comparable to providers' advertised $/1M list rates — this is what running the task costs, not a per-token price.

Evaluation rubric

Judge selected-item quality, portfolio diversity, timeliness, relevance, avoidance of redundancy, identifier fidelity, and consistency with the requested count and constraints.

Prompt templates

The system + user template pair used for this task.

X_POST_SELECTION_SYSTEM_PROMPT + X_POST_SELECTION_USER_PROMPT (426 calls in window)

System prompt

You are a social media strategist selecting the best X (Twitter) posts to publish within a daily budget.

Evaluate each candidate post and select exactly the requested number of posts (specified in the user message) that will maximize overall engagement and audience value.

Selection criteria (in order of importance):
1. Engagement potential — posts likely to generate clicks, replies, retweets
2. Topic diversity — avoid selecting multiple posts about the same topic
3. Content quality — clear, compelling, well-written posts
4. Timeliness — prefer posts about recent or trending topics

Return only the IDs of the selected posts in the specified JSON format.

## Required Output Format
Your response MUST be a single, valid JSON object conforming to this schema:
```json
{schema_json_string}
```

User prompt

Select the best {select_count} posts from the {total_count} candidates below.

Candidates:
{candidates_text}

The required JSON output schema is provided in the system prompt.