AgMoDB
ModelsAgentsOrchestratorsEvalsCompositesVisualizeIndustry
AgMoDB by @mistakeknot
All Benchmarks

BigCodeBench

Evaluates LLMs on practical, challenging coding tasks that require composing multiple function calls and libraries — closer to real-world software development than HumanEval.

codingPartial metadatabenchmark_matrix

This definition has not completed editorial and provenance review. Verify its metric, version, source, and evaluation configuration upstream before using it for a model-selection decision.

Strengths

Tests realistic multi-step coding with library usage. Higher difficulty ceiling than HumanEval.

Caveats

Relatively new benchmark with fewer evaluated models. Scores cluster in a narrow range (34–50%).

How to interpret scores

Contamination risk: moderateFreshness: periodic

Higher is better. Pass rate percentage (0–100%).

Caveat to keep in mind

Relatively new benchmark with fewer evaluated models.

Leaderboard

Last synced: Nov 22, 2026, 12:18 PM
#ModelProviderScorePrice/M
1
DeepSeek V3 (Dec '24)
source benchmark_matrix
Evidence observed Nov 22, 2026
DeepSeek50.0$0.46
2
Llama 4 Maverick
source benchmark_matrix
Evidence observed Nov 22, 2026
Meta49.7$0.42
3
GPT-4.1 mini
source benchmark_matrix
Evidence observed Nov 22, 2026
OpenAI48.9$0.70
4
Gemini 2.0 Flash (Feb '25)
source benchmark_matrix
Evidence observed Apr 21, 2026
Google45.9$0.00
5
Google: Gemini 2.0 Flash
source benchmark_matrix
Evidence observed Jul 25, 2026
Google45.9$0.17
6
Gemini 2.0 Flash (experimental)
source benchmark_matrix
Evidence observed Nov 22, 2026
Google45.9$0.00
7
Phi-4
source benchmark_matrix
Evidence observed Nov 22, 2026
Microsoft45.5$0.22
8
QwQ 32B
source benchmark_matrix
Evidence observed Nov 22, 2026
Alibaba44.6$0.00
9
Command A
source benchmark_matrix
Evidence observed Nov 22, 2026
Cohere33.8$4.38