1,140 complex Python programming tasksactiveCoding
1,140 complex Python programming tasks: BigCodeBench
Metric: Pass@1Introduced: 2024
1,140 complex Python programming tasks spanning diverse real-world domains requiring multi-library function calls. Two variants: Complete (function completion) and Instruct (natural language to code).
Models ranked
9
tracked on this benchmark
Score band
50.0 – 35.0
best → lowest tracked
Snapshot trend
-7.08
Apr 14 → Jun 7 · 3 models
Leaderboard
Tracked models ranked by Pass@1 (higher is better).
#Model variant and provenanceRelative to leaderScore
1
DeepSeek V3
50.0Version: 2025-01 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
2
Llama 4 Maverick 17B Instruct FP8
49.7Version: 2025-04 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
3
Qwen2.5-Coder-32B-Instruct
49.0Version: 2025-01 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
4
Palmyra X5
48.7Version: BigCodeBench (pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
5
GPT-4o (2024-11-20)
48.0Version: 2025-01 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
6
GPT-4o Mini (07-18)
46.1Version: 2025-01 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
7
Claude 3.5 Sonnet
44.6Version: 2025-01 (Instruct Pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Apr 14, 2026Confidence: Not recordedSource
8
Granite 4.1 30B
38.8Version: BigCodeBench (pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
9
Granite 4.1 8B
35.0Version: BigCodeBench (pass@1)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
- The model list covers the same version family you can actually deploy today.
- Top candidates overlap with your required routing and feature requirements.
Be cautious when
- There is only one benchmark snapshot or the dataset appears stale.
- The benchmark metric direction is opposite of your decision objective.
- The score difference between options is narrow and likely within implementation variance.
Related benchmarks
Last reviewed: Apr 15, 2026