# LOCA-Bench 256K

> Tests long-context agents as tool environments grow to a 256K-token environment description.

Canonical page: https://benchlm.ai/benchmarks/loca-bench-256k

- Category: [Agentic](/agentic)
- Last updated: October 29, 2026

## About LOCA-Bench 256K

- Year: 2026
- Format: Agent tasks with controlled tool-environment growth
- Paper: [LOCA-bench: Benchmarking Language Agents Under Controllable Environment Growth](https://arxiv.org/abs/2602.07962)

The 256K lane stays separate from other environment sizes and agent scaffolds. MiniMax reports M3.1 Flash at 60.3 using the official implementation. The provider does not disclose a code revision, agent strategy, reasoning effort, seeds, or run count. This provider-run result is display-only and excluded from weighted ranking.

LOCA-Bench 256K is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (1 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [MiniMax M3.1 Flash](/models/minimax-m3-1-flash) | MiniMax | 60.3 |

## FAQ

### What does LOCA-Bench 256K measure?

Tests long-context agents as tool environments grow to a 256K-token environment description.

### Which model scores highest on LOCA-Bench 256K?

MiniMax M3.1 Flash by MiniMax currently leads with a score of 60.3 on LOCA-Bench 256K.

### How many models are evaluated on LOCA-Bench 256K?

1 AI models have published results on LOCA-Bench 256K in the BenchLM catalog.

### Does LOCA-Bench 256K affect BenchLM's overall score?

Not directly. LOCA-Bench 256K is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.