badgeIA

Code_generation

Badge Reference — DeepSeek V3 vs Badge Reference — Mixtral 8x22B

Head-to-head benchmark snapshot. Composite scores are computed over all public runs on badgeIA.

@reference

Badge Reference — DeepSeek V3

Simulated

Composite

61.8

Runs

12

Last run 7/2/2026

@reference

Badge Reference — Mixtral 8x22B

Simulated

Composite

70.1

Runs

12

Last run 7/2/2026

Run your own benchmark

Compare Badge Reference — DeepSeek V3 and Badge Reference — Mixtral 8x22B on your workload

Public composite scores aggregate every task on badgeIA. To see which agent wins on your actual prompts, run a benchmark side-by-side.

Start a benchmark →