deepseek-coder-1.3B-base: benchmark results
Provider: DeepSeek. Released 2023-11-02. Access: Open.
Aggregate Rating 1158 ± 40, rank #9854 of 10916 rated models, from 12 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SAFIM | 52.63 | Average | 64.3 |
| ShaderMatch | 10.41 | Clone Match Rate (%) | 42.9 |
| EvalPlus (HumanEval+ & MBPP+) | 36.8 | Pass@1 avg (%) | 28.6 |
| C3-Bench (Controllable Code Completion) - Implementation Control | 14.2 | Instruction-Following Rate (%; 1,286 ICC tasks: passes the u | 20 |
| C3-Bench (Controllable Code Completion) - Scale Control | 4.4 | Instruction-Following Rate (%; 909 SCC tasks: AST node-type | 17.5 |
| CanItEdit (Descriptive) | 17.9 | pass@1 (%; share of the 105 hand-written Python editing prob | 12.5 |
| GSM8K | 4.4 | Accuracy (%) | 8.6 |
| CanItEdit (Lazy) | 11.76 | pass@1 (%; share of the 105 hand-written Python editing prob | 8.3 |
| C3-Bench (Controllable Code Completion) - Implementation Control - Pass@1 | 26.9 | Pass@1 (%; unit tests on the 1,286 instructed ICC tasks) | 5 |
| ARC Challenge (AI2) | 25.4 | Accuracy (%) | 2.6 |
| MMLU | 25.8 | Accuracy (%) | 1.5 |
| WinoGrande | 53.3 | Accuracy (%) | 1.2 |
Interactive version: theaggregate.ai/model?slug=deepseek-coder-1-3b-base · How It Works · Data refreshed daily, snapshot 2026-11-08.