Benchmark Almanac

Legacy-Bench

GPT-5.6 Sol 60% leads since September 2026
Not in model cards

Agents debug, add features to and migrate enterprise code in COBOL, Java 7, BASIC, C89, Fortran and Assembly. A task passes only when hidden tests all pass.

100 tasks. Factory added it to the Fireworks index on September 24th, 2026, where Fireworks runs it on 12 models.

Made by
Factory
Released
September 2026
Kind
Fixing real repositories
Level
Expert
Scored
Checked automatically
Answers
Code
Tasks
Kept private
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
September 2026
Scores refreshed
November 12th, 2026
6
Models scored on its own leaderboard

GPT-5.6 Sol leads with 60% among models released by September 2026

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

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

  1. 60.0 GPT-5.6 Sol
  2. 52.0 Claude Opus 5
  3. 51.0 GPT-6 Astra
  4. 49.0 DeepSeek V4.1 Flash
  5. 44.0 Kimi K3
  6. 23.0 GLM-5.3
  1. 1 GPT-5.6 Sol 60%
  2. 2 Claude Opus 5 52%
  3. 3 GPT-6 Astra 51%
  4. 4 DeepSeek V4.1 Flash 49%
  5. 5 Kimi K3 44%
  6. 6 GLM-5.3 23%

Source: Official Legacy-Bench leaderboard

#ModelScoreSettingDateSource
1 GPT-5.6 Sol 60% Jul 2026 Official leaderboard
2 Claude Opus 5 52% Jul 2026 Official leaderboard
3 GPT-6 Astra 51% Sep 2026 Official leaderboard
4 DeepSeek V4.1 Flash 49% Sep 2026 Official leaderboard
5 Kimi K3 44% Jul 2026 Official leaderboard
6 GLM-5.3 23% Aug 2026 Official leaderboard

Legacy-Bench score versus cost per task

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 20% 40% 60% $0$0.25$0.5$0.75$1$1.25 Cost per task ($), lower is better Pass@1 Kimi K3: Pass@1 44%, cost per task $1.09 Claude Opus 5: Pass@1 52%, cost per task $1.13 GLM-5.3: Pass@1 23%, cost per task $0.69 GPT-6 Astra: Pass@1 51%, cost per task $0.98 GPT-5.6 Sol: Pass@1 60%, cost per task $0.44 DeepSeek V4.1 Flash: Pass@1 49%, cost per task $0.12 GPT-5.6 SolDeepSeek V4.1 Flash 0% 20% 40% 60% $0$0.25$0.5$0.75$1$1.25 Cost per task ($), lower is better Pass@1 Kimi K3: Pass@1 44%, cost per task $1.09 Claude Opus 5: Pass@1 52%, cost per task $1.13 GLM-5.3: Pass@1 23%, cost per task $0.69 GPT-6 Astra: Pass@1 51%, cost per task $0.98 GPT-5.6 Sol: Pass@1 60%, cost per task $0.44 DeepSeek V4.1 Flash: Pass@1 49%, cost per task $0.12 GPT-5.6 SolDeepSeek V4.1 Flash
ModelSettingPass@1Cost per task
GPT-5.6 Sol 60% $0.44
Claude Opus 5 52% $1.13
GPT-6 Astra 51% $0.98
DeepSeek V4.1 Flash 49% $0.12
Kimi K3 44% $1.09
GLM-5.3 23% $0.69

Number one over time

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

OpenAI
Own leaderboard Not tested since Sep 2026
GPT-5.6 Sol 60%

Source: Official Legacy-Bench leaderboard

Scores over time

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

On the benchmark’s own leaderboard
0% 20% 40% 60% 80% 100% Near the ceiling (85%) GPT-5.6 Sol 0% 20% 40% 60% 80% 100% Near the ceiling (85%) GPT-5.6 Sol

Source: Official Legacy-Bench leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Coding

All 282 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