CritPt
Harbor: pendingCritPt — AI evaluation benchmark listed by opencompass.
Verified71%Knowledge & Reasoning
- Samples
- 70
- Released
- 2025
- Scores
- 138
GeoFitBench-LLM
GeoFitBench-LLM: A Benchmark for Assessing Large Language Models' Capabilities in Geospatial Data Requirements Elicitation for Geo-Analytic Tasks
A benchmark of 48 items assessing LLMs' ability to infer geospatial data requirements from natural language requests across spatio-temporal and semantic dimensions.
How to run
ltb · geofitbench-llm
$ npm i -g lettokenburn-cli$ ltb bench info geofitbench-llm
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
6 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| GPT-5.4 | 78.2 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
| Gemini 3.1 Pro | 77.4 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
| Gemini 2.5 Flash | 73.5 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
| gpt-oss-20b | 72.7 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
| DeepSeek-R1 14B | 44.7 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
| Gemma-3n | 37.1 | N/A | Independent | medium | Oct 8, 2026 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
GeoFitBench-LLM contains 48 benchmark items organized by task difficulty (D1-D3) and reasoning dimensions (Spatio-Temporal Relational, Semantic Grounding). It uses a fixed, provider-independent schema aligned with ISO standards for ground truth.
Scoring evaluates resource identification and attribute-level specification. Experiments on seven models show GPT-5.4 Thinking achieved the best result, recovering complete resources in 66.67% of requests, though fitness-for-use specification remains difficult.