BenchLeader

SWE Atlas leaderboard

As of 16 Nov 2026, the top score on SWE Atlas is 80.5% by Claude Fable 5.2 (Anthropic), ahead of Claude Opus 5.5 at 77.4%, across 28 model configurations with a published result (Scale AI SEAL, result dated 11 Nov 2026). SWE Atlas v1.1 overall: codebase QnA, test writing and refactoring on 284 tasks from 15 repositories, every model re-run with a new judge. Scale AI. SWE Atlas is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 16 Nov 2026)

  • Claude Fable 5.2 holds the top SWE Atlas v1.1 score at 80.5% (Scale AI SEAL, 11 Nov 2026).
  • Claude Opus 5.5 is second on SWE Atlas v1.1 at 77.4% (Scale AI SEAL, 9 Sep 2026).
  • SWE Atlas v1.1: SWE Atlas v1.1 overall: codebase QnA, test writing and refactoring on 284 tasks from 15 repositories, every model re-run with a new judge. Scale AI (Scale AI SEAL).
  • 28 model configurations have a published SWE Atlas v1.1 result on BenchLeader (BenchLeader, 16 Nov 2026).
  • SWE Atlas v1.1 does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

SWE Atlas v1.1 overall: codebase QnA, test writing and refactoring on 284 tasks from 15 repositories, every model re-run with a new judge. Scale AI.

Published by
Scale AI SEAL
Category
Coding
Index weight
Reference only
Models
28
Data as of
16 Nov 2026
Newest result
11 Nov 2026

Scale AI SEAL Leaderboards.

28 of 28
#
1NewClaude Fable 5.280.5%65.42026-11-11
2Claude Opus 5.577.4%61.02026-09-09
3Claude Fable 5.176.4%66.42026-09-02
4Gemini 4 Carbon71.4%–2026-11-11
5Gemini 4 Argon70.1%64.42026-06-11
6GPT-6 Astra69.4%66.42026-05-06
7Claude Sonnet 5.568.3%–2026-06-08
8GPT-6.1 Sol65.9%62.02026-05-06
9Muse Wonder64.8%64.42026-09-09
10Qwen3.8 Maxmax64.4%58.72026-11-11
11DeepSeek V4.1 Flash62.1%–2026-06-23
12Grok 4.762.1%–2026-05-06
13Inkling 2 Small61.9%57.92026-05-06
14Kimi K3.1 Pro61.1%61.82026-11-11
15Kimi K361.0%60.72026-05-06
16Claude Haiku 5.559.4%–2026-05-06
17GLM 5.358.1%–2026-11-11
18GPT-6.1 Terra57.7%–2026-05-06
19Gemini 3.8 Flash57.3%56.42026-05-06
20NewMuse Ember56.0%–2026-11-11
21GLM 5.3 Flash55.3%57.82026-11-11
22GPT-6 Luna50.8%–2026-05-06
23Mistral Large 445.2%52.92026-11-11
24Muse Spark Open42.3%–2026-05-06
25MiniMax M339.3%56.12026-11-11
26Nemotron 3.5 Super VL24.8%45.82026-11-11
27Nemotron 3 Ultra21.4%42.12026-11-11
28gpt-oss-120b8.8%46.12026-11-11

AI coding benchmark leaderboard · All LLM benchmarks

Cite as: BenchLeader, “SWE Atlas v1.1 leaderboard”, data as of 16 Nov 2026, https://www.benchleader.com/benchmarks/scale_swe_atlas.

SWE Atlas v1.1: questions

What does SWE Atlas v1.1 measure?
SWE Atlas v1.1 overall: codebase QnA, test writing and refactoring on 284 tasks from 15 repositories, every model re-run with a new judge. Scale AI. Scores are reported in percent of tasks solved; higher is better.
What is the top score on SWE Atlas?
As of 16 Nov 2026, the highest score on SWE Atlas v1.1 is 80.5% by Claude Fable 5.2 (result dated 11 Nov 2026), ahead of Claude Opus 5.5 at 77.4%. Scores come from Scale AI SEAL and are re-read daily.
Which AI model leads SWE Atlas v1.1?
Claude Fable 5.2 leads SWE Atlas v1.1 with 80.5% as of 16 Nov 2026 (result dated 11 Nov 2026), ahead of Claude Opus 5.5 at 77.4%.
How many models have SWE Atlas v1.1 results?
28 model configurations have a SWE Atlas v1.1 result on BenchLeader, all taken from Scale AI SEAL.
Who runs SWE Atlas v1.1 and how often is it updated?
SWE Atlas v1.1 is published by Scale AI SEAL. BenchLeader re-reads the published results once a day and records the date each result was published.
Does SWE Atlas v1.1 count toward the BenchLeader Index?
No. SWE Atlas v1.1 is shown for reference but left out of the composite index.