HumanEval — Results

Rank	Model	Score
1	deepseek-v4-1-flash	79.4
2	minicpm-sala	0.951
3	kimi-k2-0905	0.945
4	gpt-5-1	0.94
5	claude-3-5-sonnet	0.937
6	gpt-5-2	0.936
7	gpt-5	0.934
8	gpt-5	0.934
9	kimi-k2-instruct	0.933
10	kimi-k2-instruct	0.933
11	qwen2-5-coder-32b-instruct	0.927
12	qwen2-5-coder-32b-instruct	0.926
13	sarvam-30b	0.921
14	claude-3-5-sonnet-v1	0.92
15	mistral-large-2	0.92
16	deepseek-v3-2	0.92
17	qwen2-5-vl-32b-instruct	0.915
18	gpt-4o	0.902
19	gemini-2-5-pro	0.9
20	deepseek-v3-1	0.898
21	granite-3-3-8b-base	0.897
22	granite-3-3-8b-instruct	0.897
23	nova-2-pro	0.89
24	llama-3-1-405b-instruct	0.89
25	deepseek-v2-5	0.89

H

HumanEval — Leaderboard

1 phaseActive

HumanEval Leaderboard

No self-reported entries.