Crafter
https://github.com/danijar/crafter
Danijar Hafner's 2D Minecraft-inspired benchmark — open-world survival with 22 hierarchical achievements (collect wood → make pickaxe → mine stone → ... → collect diamond). Procedurally-generated 64x64 grid world; tests long-horizon planning, exploration, credit assignment. Lightweight (single Python file) — designed as 'inner-loop Minecraft' for fast iteration. Used as flagship benchmark for DreamerV2/V3.
At a glance
- Type
- Minecraft-inspired 2D RL benchmark with 22 achievements
- Tier
- T3
- Section
- Memory benchmarks & evaluation
- Created
- 2021-09 (Hafner arXiv 2109.06780)
- Latest release
- not applicable — not OSS
- License
- MIT
- GitHub
- 500★ Python
- Pricing
- not applicable — open benchmark; free to use
- Funding
- not applicable — not commercial
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
Lightweight enough to iterate fast (minutes per agent run); 22-achievement hierarchy gives interpretable progress signal; canonical for world-model RL research.
Cons
2D + small grid limits transfer to richer environments; only 22 achievements caps task ceiling; less brand recognition than 3D Minecraft benchmarks.
Claims & capabilities
Authored by DreamerV3's Hafner; canonical companion benchmark for world-model RL; lightweight enough to run on a single GPU.
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.
- 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.
- Atari 100k T3
Sample-efficiency benchmark protocol on Arcade Learning Environment (ALE) — agents limited to 100k environment steps (~2 hours of game-play) before evaluation. Introduced by SimPLe (Kaiser et al. 2019) and adopted by EfficientZero, IRIS, BBF, DreamerV3, Storm. Tests world-model / model-based methods where memory-as-imagination matters more than huge replay buffers.