SWE Multilingual leaderboard

A multilingual software-engineering benchmark for real-world code issue resolution across multiple programming languages.

Models ranked by their published SWE Multilingual value

Rank	Model	Provider	Repository task completion
1	Claude Opus 5	Anthropic	89.5
2	Claude Fable 5.1	Anthropic	89.1
3	Claude Opus 4.8	Anthropic	84.4
4	Hy4 preview	Tencent	82.9
5	Qwen3.8-Flash-Next	Alibaba	81
6	Qwen3.8-Omni-Flash	Alibaba	80.5
7	Composer 2.5	Cursor	79.8
8	OAOrnith-1.5-397B	Ornith AI	79.6
9	DAOrnith-1.0-397B	DeepReinforce AI	78.9
10	Laguna S 2.1	Poolside	78.5
11	Claude Sonnet 5	Anthropic	78.3
11	Qwen3.7 Max	Alibaba	78.3
13	Grok 4.5	xAI	78
14	SWE-1.7	Cognition	77.8
15	Claude Opus 4.5	Anthropic	77.5
16	Kimi K2.6	Moonshot AI	76.7
17	MiniMax M2.7	MiniMax	76.5
18	DeepSeek V4 Pro 0813	DeepSeek	76.2
19	Qwen3.7 Plus	Alibaba	75.8
20	DSdots3-note Preview	Dots Studio	75.7
21	Qwen3.6 Plus	Alibaba	73.8
22	Composer 2	Cursor	73.7
23	DeepSeek V4 Flash 0731	DeepSeek	73.3
23	GLM-5	Z.AI	73.3
25	Kimi K2.5	Moonshot AI	73
26	Ling 3.0 Flash	InclusionAI	72.4
27	OAOrnith-1.5-35B-A3B	Ornith AI	71.4
28	Qwen3.6-27B	Alibaba	71.3
29	DAOrnith-1.0-35B	DeepReinforce AI	69.3
30	Nemotron 3 Ultra	NVIDIA	67.7
31	Qwen3.6-35B-A3B	Alibaba	67.2
32	Laguna M.1	Poolside	63.1
32	Laguna XS 2.1	Poolside	63.1
34	LongCat-Flash-Lite-Sparse	Meituan	59.33
35	Laguna XS.2	Poolside	57.7
36	OAOrnith-1.5-9B	Ornith AI	54.4
37	DAOrnith-1.0-9B	DeepReinforce AI	52
38	Granite 4.2 30B	IBM	41.89
39	Nemotron 3.5 Lightning 30B A3B NVFP4	NVIDIA	36.47
40	Granite 4.2 8B	IBM	30.78
41	LLaDA2.2-flash	InclusionAI	25

Evidence key: Observed

Rows are ordered by the value MiniMax M2.7: Early Echoes of Self-Evolution published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.