LatchBio SpatialBench
Explore score history ↗159 spatial transcriptomics analysis evaluations, each run 3 times and averaged; the published 95% intervals span about ±6 to 7 points, so neighbouring models usually overlap. The score depends on the agent harness as well as the model (Claude Opus 4.8 scores 55.35 in Claude Code and 52.62 in mini-swe-agent), and the evaluation container has network access. The arXiv paper (2512.21907) reports an earlier task set, so its numbers are not comparable with these.
- Evaluates
- biology
- Openness
- unknown
- Importer
- Model reports
LatchBio SpatialBench code
LatchBio SpatialBench results and reported scores
Reported benchmark scores · Model reports
Reported scores: 20
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Claude Opus 4.5 | Anthropic | 42.77 | 2026-11-01 | evidence |
| Claude Opus 4.6 | Anthropic | 52.83 | 2026-11-01 | evidence |
| Claude Opus 4.7 | Anthropic | 51.36 | 2026-11-01 | evidence |
| Claude Opus 4.7 | Anthropic | 52.41 | 2026-11-01 | evidence |
| Claude Opus 4.8 | Anthropic | 55.35 | 2026-11-01 | evidence |
| Claude Opus 4.8 | Anthropic | 52.62 | 2026-11-01 | evidence |
| Claude Sonnet 4.5 | Anthropic | 41.51 | 2026-11-01 | evidence |
| Claude Sonnet 4.6 | Anthropic | 44.23 | 2026-11-01 | evidence |
| Gemini 2.5 Pro | 28.93 | 2026-11-01 | evidence | |
| Gemini 3.1 Pro Preview | 51.57 | 2026-11-01 | evidence | |
| Gemini 3.5 Flash | 48.85 | 2026-11-01 | evidence | |
| Gemini 3.5 Flash | 55.56 | 2026-11-01 | evidence | |
| GPT-5.1 | OpenAI | 39.83 | 2026-11-01 | evidence |
| GPT-5.2 | OpenAI | 50.1 | 2026-11-01 | evidence |
| GPT-5.4 | OpenAI | 57.44 | 2026-11-01 | evidence |
| GPT-5.5 | OpenAI | 53.67 | 2026-11-01 | evidence |
| GPT-5.5 | OpenAI | 57.65 | 2026-11-01 | evidence |
| Grok 4 | xAI | 31.87 | 2026-11-01 | evidence |
| Grok 4.1 Fast (reasoning) | xAI | 33.96 | 2026-11-01 | evidence |
| Grok 4.20 Beta 0309 (reasoning) | xAI | 45.91 | 2026-11-01 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.
Which sources cite LatchBio SpatialBench?
- SpatialBench results table (latchbio/spatialbench, commit baa9841) LatchBio, 2026-11-01