What is the best open LLM on FrontierSWE?

GLM 5.3 is the top open model on FrontierSWE, scoring 30.2%. Among all models tested — including proprietary ones — it ranks #9. The top model overall is GPT 6 Astra Max (OpenAI) at 65.5%.

Can open models match proprietary models on FrontierSWE?

Not quite on FrontierSWE: the strongest proprietary model (GPT 6 Astra Max) scores 65.5%, ahead of the best open model (GLM 5.3) at 30.2% — but you can run the open one yourself.

Coding

FrontierSWE Leaderboard

Name: FrontierSWE — open LLM scores
Creator: epoch

FrontierSWE, from Proximal, tests coding agents on long-horizon software-engineering tasks spanning implementation, performance optimisation and open-ended research, each run for hours through an agent harness. The score is the average normalised task score across repeated runs of every task.

Source: epoch4 open models ranked+15 proprietaryData through Sep 2026

Open models All models

All models ranked on FrontierSWE

Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.

#	Model	Score
1	GPT 6 Astra Max · proprietary	65.5%
2	Claude Opus 5.5 Max · proprietary	62.3%
3	Claude Sonnet 5.5 Max · proprietary	61.9%
4	Claude Fable 5.1 Max · proprietary	56.3%
5	Gemini 4 Argon (high) · proprietary	55.0%
6	Claude Opus 5 Max · proprietary	52.0%
7	Claude Fable 5 Max · proprietary	47.0%
8	GPT 5.6 Sol Max · proprietary	32.2%
9	GLM 5.3 · 753.3B	30.2%
10	Grok 4.7 (xhigh) · proprietary	29.5%
11	Kimi K3 · 2779.9B	25.9%
12	Grok 4.6 (xhigh) · proprietary	25.3%
13	Gemini 3.7 Flash (high) · proprietary	20.3%
14	Gemini 3.8 Flash (high) · proprietary	19.6%
15	GLM 5.3 Flash · 321.3B	18.1%
16	Qwen3.8 Max (Sep 02, xhigh) · proprietary	17.8%
17	Qwen3.8 Max (xhigh) · proprietary	15.8%
18	Muse Spark 1.2 (xhigh) · proprietary	12.0%
19	Inkling · 952.4B	4.1%

FrontierSWE: frequently asked questions

What is the best open LLM on FrontierSWE?: GLM 5.3 is the top open model on FrontierSWE, scoring 30.2%. Among all models tested — including proprietary ones — it ranks #9. The top model overall is GPT 6 Astra Max (OpenAI) at 65.5%.
Can open models match proprietary models on FrontierSWE?: Not quite on FrontierSWE: the strongest proprietary model (GPT 6 Astra Max) scores 65.5%, ahead of the best open model (GLM 5.3) at 30.2% — but you can run the open one yourself.

Scores aggregated from epoch. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.