Benchmark Almanac

Cognition CodeSearch Eval

SWE-grep 0.66 tops its leaderboard GPT-5 0.59 leads the models we track
Not in model cards

Agents retrieve relevant files and line ranges for questions about real repositories. Cognition compares their answers with human labels using precision-weighted F scores.

The SWE-grep launch compares both in-house models with Claude Opus, Sonnet, Haiku and multiple GPT models. It also reports retrieval latency.

Made by
Cognition
Released
October 2025
Kind
Fixing real repositories
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Agentic
Environment
Terminal
Items from
Real work and use
Refresh
One fixed set
Run by
A host
In views
Agents
Strongest model tested
GPT-5 #71 on the capability index
Last model added
October 2025
Scores refreshed
November 21st, 2026
8
Entries in its paper, 6 models we track

SWE-grep tops the leaderboard with 0.66 3 models share the top score of 0.59 among models released by October 2025

Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is GPT-5, at 0.59. This view shows only models we track, each at its best.

The newest model with a score here came out in October 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 SWE-grep 0.66
  2. 2 GPT-5 High 0.59
  3. 3 GPT-5-Codex 0.59
  4. 4 Claude Sonnet 4.5 non-thinking 0.59
  5. 5 SWE-grep-mini 0.58
  6. 6 Claude Haiku 4.5 0.56
  7. 7 Claude Opus 4.1 interleaved thinking 0.55
  8. 8 GPT-5 mini Minimal 0.36

Source: The Cognition CodeSearch Eval paper

#EntryScoreSettingDateSource
1 SWE-grep 0.66 Official leaderboard
2 GPT-5 0.59 High Aug 2025 Official leaderboard
3 GPT-5-Codex 0.59 Sep 2025 Official leaderboard
4 Claude Sonnet 4.5 0.59 non-thinking Sep 2025 Official leaderboard
5 SWE-grep-mini 0.58 Official leaderboard
6 Claude Haiku 4.5 0.56 Oct 2025 Official leaderboard
7 Claude Opus 4.1 0.55 interleaved thinking Aug 2025 Official leaderboard
8 GPT-5 mini 0.36 Minimal Aug 2025 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5 0.59 High Aug 2025 Official leaderboard
2 GPT-5-Codex 0.59 Sep 2025 Official leaderboard
3 Claude Sonnet 4.5 0.59 non-thinking Sep 2025 Official leaderboard
4 Claude Haiku 4.5 0.56 Oct 2025 Official leaderboard
5 Claude Opus 4.1 0.55 interleaved thinking Aug 2025 Official leaderboard
6 GPT-5 mini 0.36 Minimal Aug 2025 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5
Minimal Low Medium High
0.59
Claude Sonnet 4.5
non-thinking
0.59
Claude Opus 4.1
non-thinking
0.55

Source: The Cognition CodeSearch Eval paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5 High 0.59
GPT-5-Codex 0.59
Claude Sonnet 4.5 non-thinking 0.59
Claude Sonnet 4.5 interleaved thinking 0.59
GPT-5 Medium 0.56
Claude Haiku 4.5 0.56
Claude Opus 4.1 interleaved thinking 0.55
Claude Opus 4.1 non-thinking 0.54
GPT-5 Low 0.54
GPT-5 Minimal 0.45
GPT-5 mini Minimal 0.36

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5
Minimal Low Medium High
0.59
Claude Sonnet 4.5
non-thinking
0.59
Claude Opus 4.1
non-thinking
0.55

Source: The Cognition CodeSearch Eval paper

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5 High 0.59
GPT-5-Codex 0.59
Claude Sonnet 4.5 non-thinking 0.59
Claude Sonnet 4.5 interleaved thinking 0.59
GPT-5 Medium 0.56
Claude Haiku 4.5 0.56
Claude Opus 4.1 interleaved thinking 0.55
Claude Opus 4.1 non-thinking 0.54
GPT-5 Low 0.54
GPT-5 Minimal 0.45
GPT-5 mini Minimal 0.36

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 Oct 2025
GPT-5 0.59

Source: The Cognition CodeSearch Eval 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
0.40 0.50 0.60 0.40 0.50 0.60

Source: Official Cognition CodeSearch Eval 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 13 months ago, in October 2025, for Claude Haiku 4.5. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (8 entries).
  • Pass Models scored 6 models we track, plus 2 other entries on its leaderboard.
  • Warning Newest score The newest came 13 months ago, in October 2025, for Claude Haiku 4.5.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5 (#71).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on cognition.com: 11 rows for models we track and 2 other entries at the last read.
  • Pass Last read November 21st, 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