SmartPlay
https://github.com/Microsoft/SmartPlay
Microsoft Research's SmartPlay — LLM-agent benchmark on 6 games (Bandits, Rock-Paper-Scissors, Tower of Hanoi, MessengerEnv, Crafter, Minecraft). Tests 9 distinct capability axes including object understanding, planning, generalisation, error-handling, theory-of-mind. ICLR 2024. Bridges narrow-task benchmarks (MMLU) and full-agent benchmarks (AgentBench).
Archived · Verified 2026-10-28 · fresh as of 2026-10-28 · 30 cited facts, 0 with a third-party source · 1 recorded change · Suggest a correction
At a glance
- Tier
- T3
- Section
- Memory benchmarks & evaluation
- Created
- 2023-10 (Wu et al. arXiv 2310.01557) repo fix
- Latest release
- not applicable — not OSS fix
- GitHub
- 170★ Python repo fix
- Pricing
- not applicable — open benchmark; free to use repo fix
- Funding
- not applicable — not commercial fix
- Freshness
- 2026-10-28 fix
- Layer
- agent-infra repo fix
- Licensing
- open-source repo 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
9 capability axes give finer-grained agent capability picture than single-task benchmarks; lightweight (LLM-interface, not full RL).
Cons
Text-only LLM interface bypasses perception; 6-game set is narrow; ICLR 2024 already feels dated vs newer agent benchmarks.
Claims & capabilities
ICLR 2024; 9 capability axes evaluated across 6 games; GPT-4 SOTA at time of release
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 (29 cited facts, 0 third-party)
Each fact links to the page it came from; the badge says who published it (how to read sources).
- What it is LLM-agent benchmark across 6 games — Microsoft Research repo
- What it does & what's distinct Microsoft Research's SmartPlay — LLM-agent benchmark on 6 games (Bandits, Rock-Paper-Sci… repo
- Claims / benefits ICLR 2024; 9 capability axes evaluated across 6 games; GPT-4 SOTA at time of release repo
- Modalities text-converted game state (LLM-interfacing layer) repo
- Created 2023-10 (Wu et al. arXiv 2310.01557) repo
- License MIT repo
- GitHub 170★ Python repo
- Mindshare Microsoft Research; one of the few systematic LLM-agent-vs-games benchmarks repo
- Citations 80+ cites — Wu et al. ICLR 2024 repo
- Pricing not applicable — open benchmark; free to use repo
- Deployment not applicable — run locally repo
- Founders / pedigree Yue Wu, Xuan Tang, Tom M. Mitchell, Yuanzhi Li (Microsoft Research / CMU) repo
- Reproducibility Independently reproduced — multiple LLM-agent papers cite results repo
- Code/weights release Code public repo
- Versioning not applicable — static benchmark repo
- Tombstoning not applicable — static benchmark repo
- Schema evolution not applicable — fixed evaluation dataset repo
- Forgetting policy not applicable — static benchmark repo
- Pros 9 capability axes give finer-grained agent capability picture than single-task benchmark… repo
- Cons Text-only LLM interface bypasses perception; 6-game set is narrow; ICLR 2024 already fee… repo
- Project & sources github.com/Microsoft/SmartPlay arxiv 2310.01557 repo
- commit-trajectory [{"ym":"2023-10","cum":33},{"ym":"2023-11","cum":48},{"ym":"2023-12","cum":56},{"ym":"20… repo
- cost-input-usd-per-mtok 0 repo
- cost-output-usd-per-mtok 0 repo
- cost-tier free repo
- cost-pricing-model free-tier repo
- cost-last-verified 2026-06-29 repo
- licensing-bucket open-source repo
- taxonomy-three-layer-rollup agent-infra repo
What changed
Changes to this record's lifecycle, licence, release, pricing and hosting facts, from the catalog's git history. Subscribe (Atom)
- 2026-10-22
- Lifecycle first recorded as “archived”