Skip to content
Weights Not Walls
Menu

FrontierSWE

Reported

No description yet: we only describe an eval set when we can do it correctly.

Kind
Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
Link
https://www.frontierswe.com/
Score shown
Epoch's column "Score", as a percentage, higher is better; random guessing scores about 0.0%. The source gives no error bars.
Results
25 in all: 5 for open-weight models, 20 for closed ones. 5 linked to a model page here.

Best open-weight result: zai-org/GLM-5.3, 30.2%, number 13 of 25 results.

FrontierSWE results
#ModelWeightsSettingScoreDateSource
1claude-fable-5-2_maxAnthropicclosedeffort max; Harness: proximus69.9%not datedfrontierswe.com
2gpt-6-astra_maxOpenAIclosedeffort max; Harness: proximus65.5%not datedfrontierswe.com
3claude-opus-5-5_maxAnthropicclosedeffort max; Harness: proximus62.3%not datedfrontierswe.com
4claude-sonnet-5-5_maxAnthropicclosedeffort max; Harness: proximus61.9%not datedfrontierswe.com
5gemini-4-carbon_highGoogle DeepMindclosedeffort high; Harness: proximus61.2%not datedfrontierswe.com
6claude-fable-5-1_maxAnthropicclosedeffort max; Harness: proximus56.3%not datedfrontierswe.com
7gemini-4-argon_highGoogle DeepMindclosedeffort high; Harness: proximus55.0%not datedfrontierswe.com
8gemini-4-hydrogen_highGoogle DeepMindclosedeffort high; Harness: proximus53.1%not datedfrontierswe.com
9claude-opus-5_maxAnthropicclosedeffort max; Harness: proximus52.0%not datedfrontierswe.com
10claude-fable-5_maxAnthropicclosedeffort max; Harness: proximus47.0%not datedfrontierswe.com
11claude-haiku-5-5_maxAnthropicclosedeffort max; Harness: proximus43.8%not datedfrontierswe.com
12gpt-5.6-sol_maxOpenAIclosedeffort max; Harness: proximus32.2%not datedfrontierswe.com
13zai-org/GLM-5.3Epoch: glm-5.3_max, Z.ai (Zhipu AI)openeffort max; Harness: proximus30.2%not datedfrontierswe.com
14grok-4.7_xhighxAIclosedeffort xhigh; Harness: proximus29.5%not datedfrontierswe.com
15moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshotopeneffort max; Harness: proximus25.9%not datedfrontierswe.com
16muse-spark-1.3_maxMeta AIclosedeffort max; Harness: proximus25.8%not datedfrontierswe.com
17grok-4.6_xhighxAIclosedeffort xhigh; Harness: proximus25.3%not datedfrontierswe.com
18thinkingmachines/Inkling-2-SmallEpoch: inkling-2-small_xhigh, Thinking Machinesopeneffort xhigh; Harness: proximus25.1%not datedfrontierswe.com
19gemini-3.7-flash_highGoogle DeepMindclosedeffort high; Harness: proximus20.3%not datedfrontierswe.com
20gemini-3.8-flash_highGoogle DeepMindclosedeffort high; Harness: proximus19.6%not datedfrontierswe.com
21zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI)openeffort max; Harness: proximus18.1%not datedfrontierswe.com
22qwen3.8-max-0902_xhighAlibabaclosedeffort xhigh; Harness: proximus17.8%not datedfrontierswe.com
23qwen3.8-max_xhighAlibabaclosedeffort xhigh; Harness: proximus15.8%not datedfrontierswe.com
24muse-spark-1.2_xhighMeta AIclosedeffort xhigh; Harness: proximus12.0%not datedfrontierswe.com
25thinkingmachines/InklingEpoch: Inkling_xhigh, Thinking Machinesopeneffort xhigh; Harness: proximus4.1%not datedfrontierswe.com

Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method