Best LLMs for Research Region Identification

Category: Structured Data & Fact Extraction · Typical I/O: 1263→518 tokens

Runs on model group: All models. Inputs may reach any model in the group, not only the ranked models below.

Run this task on Fronset — request an invitation

Models

Frontier on this task: Moonshot Kimi K3 at 8.84 / 10. Quality bar at 90%: 7.95.

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 does not clear the bar.

Model	Quality score	CI low	Cost / 1k runs	vs best value
Thinking Machines Inkling Small	8.50 / 10	8.29	$4.44	best value
Thinking Machines Inkling	8.39 / 10	8.17	$13.77	3.1x more expensive
Moonshot Kimi K3	8.84 / 10	8.63	$13.84	3.1x more expensive
Gemini 3.5 Flash Lite	5.32 / 10	4.97	$0.40	91% cheaper

Cost breakdown

Model	Quality	Confidence	Cost / 1k runs	Overpay	Mode
Thinking Machines Inkling Small ★ OpenRouter	8.50 / 10 CI [8.29, 8.72]	HIGH	$4.44	best value	batch
Thinking Machines Inkling OpenRouter	8.39 / 10 CI [8.17, 8.60]	HIGH	$13.77	3.1x	batch
Moonshot Kimi K3 best Moonshot AI	8.84 / 10 CI [8.63, 9.04]	HIGH	$13.84	3.1x	batch

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: 1263 input tokens → 518 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 evidence-based regional relevance, ranking, coverage versus focus trade-off, allowed-set compliance, and restraint against unsupported regions. Cardinality and identifier validity are deterministic.

Output schema

Every answer on this task is checked against this JSON Schema, whichever model wrote it. An answer that doesn't fit counts as a model failure, and the call is retried on another model.

{
  "description": "Output model for region identification",
  "properties": {
    "reasoning": {
      "default": "",
      "description": "Explanation for region selection",
      "title": "Reasoning",
      "type": "string"
    },
    "region_codes": {
      "description": "List of relevant region codes",
      "items": {
        "type": "string"
      },
      "title": "Region Codes",
      "type": "array"
    }
  },
  "required": [
    "region_codes"
  ],
  "title": "RegionIdentificationOutput",
  "type": "object"
}

Prompt templates

The published template for this task.

default · LLMB_RESEARCH_REGION_IDENTIFICATION_SYSTEM + LLMB_RESEARCH_REGION_IDENTIFICATION_USER (0 calls in window)

System prompt

Rank regions by their likely contribution of authoritative, decision-relevant evidence for the stated subject and research purpose. Use only the supplied region set. Distinguish headquarters, operating footprint, market importance, and regulatory jurisdiction, and do not infer a global scope from generic wording. Select multiple regions only when each adds material evidence coverage. Treat empty optional values as absent and return only the requested result. Your response must conform exactly to this output schema: {schema_json_string}.

User prompt

Inputs — subject_name: {subject_name}; subject_code: {subject_code}; subject_type: {subject_type}; subject_description: {subject_description}; available_regions: {available_regions}; additional_context: {additional_context}. Use only these inputs to complete the task defined by the system prompt.