Benchmark Almanac

SWE-Marathon-Ext

GPT-5.6 Sol 74% leads since August 2026
Not in model cards

Agents build 12 software products from specifications. Automated tests and browser checks grade the applications.

Mercor publishes a leaderboard for eight frontier models. Each model completed three independent runs per task.

Made by
Mercor
Released
August 2026
Kind
Building whole projects
Level
Expert
Scored
Checked automatically
Answers
Agent task
Measures
Capability
Language
English
In and out
Text → Actions, Code
Interaction
Agentic
Environment
Terminal
Refresh
One fixed set
Run by
A host
Based on
SWE-Marathon
In views
Agents
Strongest model tested
GPT-5.6 Sol #14 on the capability index
Last model added
July 2026
Scores refreshed
November 20th, 2026
8
Models scored in its paper

GPT-5.6 Sol leads with 74% among models released by July 2026

Best score per model from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in July 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.6 Sol 74%
  2. 2 GPT-5.5 73%
  3. 3 Grok 4.5 67%
  4. 4 Claude Opus 4.8 65%
  5. 5 Gemini 3.5 Flash 42%
  6. 6 Gemini 3.1 Pro 34%
  7. 7 GPT-5.6 Luna 28%
  8. 8 GLM-5.2 27%

Source: The SWE-Marathon-Ext paper

#ModelScoreSettingDateSource
1 GPT-5.6 Sol 74% Terminus on Harbor Jul 2026 Official leaderboard
2 GPT-5.5 73% Terminus on Harbor Apr 2026 Official leaderboard
3 Grok 4.5 67% Terminus on Harbor Jul 2026 Official leaderboard
4 Claude Opus 4.8 65% Terminus on Harbor May 2026 Official leaderboard
5 Gemini 3.5 Flash 42% Terminus on Harbor May 2026 Official leaderboard
6 Gemini 3.1 Pro 34% Terminus on Harbor Feb 2026 Official leaderboard
7 GPT-5.6 Luna 28% Terminus on Harbor Jul 2026 Official leaderboard
8 GLM-5.2 27% Terminus on Harbor Jun 2026 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s paper: Deterministic backend API and Computer-use agent browser interface. Each ranks models on one part of the test. GPT-5.5 leads Deterministic backend API and GPT-5.6 Sol leads Computer-use agent browser interface. Arrows show how far a model moved from its overall rank among the models we track.

Deterministic backend API

8 models we track have a score here.

  1. 1 GPT-5.5 ↑1 78%
  2. 2 Claude Opus 4.8 ↑2 71%
  3. 3 GPT-5.6 Sol ↓2 71%
  4. 4 Grok 4.5 ↓1 70%
  5. 5 Gemini 3.5 Flash 43%
  6. 6 Gemini 3.1 Pro 40%
  7. 7 GPT-5.6 Luna 39%
  8. 8 GLM-5.2 33%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Benchmark paper Not tested since Aug 2026

Source: The SWE-Marathon-Ext paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.6 Sol GPT-5.5 Gemini 3.1 Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.6 Sol Gemini 3.1 Pro

Source: Official SWE-Marathon-Ext leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

9 of 10 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (8 entries).
  • Pass Models scored 8 models we track.
  • Pass Newest score The newest is from July 2026, for GPT-5.6 Luna.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.6 Sol (#14).
  • Pass Headroom The best is 74%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on mercor.com: 8 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check