Catalog Systems Memory benchmarks & evaluation BALROG

BALROG

https://balrogai.com/

BALROG (Benchmarking Agentic LLM and VLM Reasoning on Games) — 2024 benchmark suite covering NetHack, BabyAI, Crafter, MiniHack, TextWorld, Baba Is AI. Specifically targets long-horizon, partial-observability, exploration-heavy games where memory and planning matter. SOTA GPT-4o and Claude 3.5 Sonnet both score <25% on aggregate. Released by AI Safety Institute UK + Cambridge collaborators.

Active · Verified 2026-10-28 · fresh as of 2026-10-28 · 29 cited facts, 0 with a third-party source · Suggest a correction

At a glance

Type
LLM/VLM agent benchmark on 6 challenging game environments vendor fix
Tier
T4
Section
Memory benchmarks & evaluation
Created
2024-11 (Paglieri et al. arXiv 2411.13543) vendor fix
Latest release
not applicable — not OSS fix
License
MIT vendor fix
GitHub
200★ Python vendor fix
Pricing
not applicable — open benchmark; free to use vendor fix
Funding
not applicable — not commercial fix
Freshness
2026-10-28 fix
Layer
agent-infra repo fix
Licensing
open-source vendor fix

Memory taxonomy

storage
n/a
retrieval
n/a
persistence
n/a
update
read-only
unit
n/a
governance
n/a
conflict
n/a

When to use

Optimised for: not applicable - eval dataset, not a system

Anti-fit: not applicable - eval dataset, not a system

Pros & cons

Pros

Specifically designed for memory + long-horizon + partial-observability evaluation; active leaderboard tracks frontier models monthly; six diverse environments.

Cons

Recent (2024-11) — limited follow-up adoption yet; aggregate SOTA still very low → high noise in evaluation.

Claims & capabilities

2024 ICLR-track preprint; SOTA <25% — frontier models all struggle on long-horizon games; 6 environments selected for memory / planning / partial-observability

Technical surface

API surface
not applicable — eval dataset, not a system
Backend storage
not applicable — eval dataset, not a system
Deployment
not applicable — run locally
Embedding model
not applicable — eval dataset, not a system
Multi-tenancy
not applicable — eval dataset, not a system
MCP
not applicable — benchmark / evaluation harness
A2A
not applicable — benchmark / evaluation harness
OpenTelemetry
not applicable — benchmark / evaluation harness

Similar systems

Other memory benchmarks & evaluation in the catalog, ranked by inbound references.

  • LoCoMo T3

    Conversations spanning up to 35 sessions / 300 turns / ~9K tokens, built from LLM-agent personas grounded on temporal event graphs and verified by human annotators. Tests QA (single-hop, multi-hop, temporal, commonsense, adversarial), event-graph summarisation, multimodal dialogue.

  • GAIA T3

    Benchmark for general AI assistants made of real-world questions that require reasoning, multimodality, web browsing and tool use, with a public leaderboard on Hugging Face. Traction: ~917 likes on the Hugging Face dataset

  • LongBench T3

    Tsinghua THUDM. 21 datasets across 6 categories (single-doc QA, multi-doc QA, summarisation, few-shot ICL, synthetic, code completion) in English + Chinese. ~6.7K words English / 13.4K chars Chinese average. v2 at ACL 2025 with harder, longer instances.

  • LongMemEval T3

    5 core memory abilities — info extraction, multi-session reasoning, temporal reasoning, knowledge updates, abstention — via 500 curated questions in synthetic chat histories. Two scales: LongMemEval-S (115K tokens), LongMemEval-M (up to 1.5M).

  • RULER T3

    NVIDIA. Extends NIAH with multi-needle retrieval, multi-hop entity tracing, aggregation, QA — configurable for arbitrary length / difficulty. Evaluated 17 long-context models on 13 tasks.

  • InfiniteBench (∞Bench) T3

    Tsinghua / OpenBMB. First benchmark with average input >100K tokens. 12 tasks across English + Chinese: long novel QA, code debugging, math in long context, summarisation, retrieval. Realistic documents (novels, code) rather than synthetic filler.

Sources (28 cited facts, 0 third-party)

Each fact links to the page it came from; the badge says who published it (how to read sources).