Developer

Autonomous Coding Agent

End-to-end autonomous software engineering: reading issues, writing code, running tests, submitting PRs.

task.agentic_multi_step_completiontask.code_generation

Evidence quality is currently limited for this use case. Rankings below are useful for exploration, not a strong winner claim.

Provisional leader

claude-opus-4-6

Current leader based on limited benchmark evidence. Treat this ranking as directional until coverage improves.

28.4%

Best benchmark score

30.9%

Confidence

All ranked models — top 3

🥇

claude-opus-4-6

28.4%

🥈

gpt-5-2025-08-07

24.9%

🥉

claude-sonnet-4

22.0%

Ranked Models

Evidence Quality

82%

Evidence Points

Top Signal

SWE-bench Verified Leaderboard: swe_verified_resolved_pct

All Ranked Models

30 of 30 models

Rank	Model	Score	Confidence	Price / 1M	Evidence sources
🥇	claude-opus-4-6 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and OpenHands Index average_score_pct	28.4%	31%	$10.00	SWE-bench Verified LeaderboardOpenHands Index
🥈	gpt-5-2025-08-07 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	24.9%	29%	—	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#4	claude-sonnet-4 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	22.0%	29%	$6.00	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#9	GLM-5 Strong on OpenHands Issue Resolution issue_resolution_score_pct and Sonar Java Quality Leaderboard functional_skill_pct	15.9%	28%	—	OpenHands Issue ResolutionSonar Java Quality Leaderboard
#10	Kimi K2 Thinking Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Sonar Java Quality Leaderboard functional_skill_pct	15.9%	41%	$1.07	SWE-bench Verified LeaderboardSonar Java Quality Leaderboard
#11	claude-sonnet-4.6 Strong on OpenHands Issue Resolution issue_resolution_score_pct and OpenHands Index greenfield_score_pct	15.8%	30%	$6.00	OpenHands Issue ResolutionOpenHands Index
#13	o3-20250416 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	15.2%	20%	$3.50	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#15	gemini-3-pro-preview Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Vals SWE-bench overall_accuracy_pct	14.5%	18%	$4.50	SWE-bench Verified LeaderboardVals SWE-bench
#16	gemini-2.5-pro Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Galileo Agent Leaderboard v2 Avg AC	14.3%	20%	$3.44	SWE-bench Verified LeaderboardGalileo Agent Leaderboard v2
#18	gpt-4o-2024-05-13 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and RepoQA Official Results overall_average_pass_at_1_pct	14.2%	20%	—	SWE-bench Verified LeaderboardRepoQA Official Results
#19	claude-opus-4 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	13.4%	16%	$10.00	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#20	gpt-4.1-20250414 Strong on Galileo Agent Leaderboard v2 Avg AC and SWE-bench Verified Leaderboard swe_verified_resolved_pct	13.3%	21%	—	Galileo Agent Leaderboard v2SWE-bench Verified Leaderboard
#21	gpt-4.1 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	13.3%	15%	$3.50	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#22	gpt-4o-20241120 Strong on BigCodeBench Official bigcodebench_complete_pct and SWE-bench Verified Leaderboard swe_verified_resolved_pct	13.3%	25%	—	BigCodeBench OfficialSWE-bench Verified Leaderboard
#23	kimi-k2.5-thinking Strong on OpenHands Index average_score_pct and Vals LiveCodeBench overall_accuracy_pct	13.2%	30%	—	OpenHands IndexVals LiveCodeBench
#24	gpt-5.2-2025-12-11 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Vals SWE-bench overall_accuracy_pct	13.1%	15%	—	SWE-bench Verified LeaderboardVals SWE-bench
#27	o4-mini Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Aider Polyglot Leaderboard percent_correct_pct	12.2%	19%	$1.93	SWE-bench Verified LeaderboardAider Polyglot Leaderboard
#30	claude-opus-4-5-20251101 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Vals Terminal-Bench 2 overall_accuracy_pct	11.9%	14%	—	SWE-bench Verified LeaderboardVals Terminal-Bench 2
#31	deepseek-r1 Strong on Aider Polyglot Leaderboard percent_correct_pct and Sonar Java Quality Leaderboard functional_skill_pct	11.5%	16%	$0.27	Aider Polyglot LeaderboardSonar Java Quality Leaderboard
#32	gpt-5-mini-2025-08-07 Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Vals LiveCodeBench overall_accuracy_pct	11.1%	16%	—	SWE-bench Verified LeaderboardVals LiveCodeBench
#33	qwen-2.5-72b-instruct Strong on Galileo Agent Leaderboard v2 Avg AC and BigCodeBench Official bigcodebench_complete_pct	11.1%	15%	—	Galileo Agent Leaderboard v2BigCodeBench Official
#34	minimax-m2.1 Strong on Sonar Java Quality Leaderboard functional_skill_pct and Vals SWE-bench overall_accuracy_pct	10.7%	34%	$0.53	Sonar Java Quality LeaderboardVals SWE-bench
#36	gpt-4o Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and τ-bench Airline (Official README) tau_airline_pass1_pct	10.5%	21%	$0.26	SWE-bench Verified Leaderboardτ-bench Airline (Official README)
#44	Grok-4-0709 Strong on Galileo Agent Leaderboard v2 Avg AC and Galileo Agent Leaderboard v2 Avg TSQ	8.9%	14%	—	Galileo Agent Leaderboard v2Galileo Agent Leaderboard v2
#45	deepseek-v3 Strong on BigCodeBench Official bigcodebench_complete_pct and Galileo Agent Leaderboard v2 Avg AC	8.8%	12%	—	BigCodeBench OfficialGalileo Agent Leaderboard v2
#46	gpt-4o-2024-08-06 Strong on BigCodeBench Official bigcodebench_hard_complete_pct and Aider Code Editing Leaderboard percent_correct_pct	8.8%	18%	—	BigCodeBench OfficialAider Code Editing Leaderboard
#47	Kimi-K2-Instruct Strong on SWE-bench Verified Leaderboard swe_verified_resolved_pct and Galileo Agent Leaderboard v2 Avg AC	8.7%	13%	—	SWE-bench Verified LeaderboardGalileo Agent Leaderboard v2
#49	glm-4.7 Strong on Sonar Java Quality Leaderboard functional_skill_pct and Vals LiveCodeBench overall_accuracy_pct	8.0%	12%	—	Sonar Java Quality LeaderboardVals LiveCodeBench
#51	gpt-4.1-mini-20250414 Strong on Galileo Agent Leaderboard v2 Avg AC and SWE-bench Verified Leaderboard swe_verified_resolved_pct	7.9%	15%	—	Galileo Agent Leaderboard v2SWE-bench Verified Leaderboard
#54	Llama 3.3 70B Instruct Strong on BigCodeBench Official bigcodebench_complete_pct and BigCodeBench Official bigcodebench_instruct_pct	7.0%	14%	—	BigCodeBench OfficialBigCodeBench Official

Compare Models

Select two different models above to compare their evidence side by side.

▶Ranking diagnostics & missing models

Source lift

Ranked

Sources

Quality

Low

SWE-bench Verified Leaderboard

19 rows · 3.5% avg lift

Vals LiveCodeBench

19 rows · 0.8% avg lift

Vals MedQA

14 rows · 0.2% avg lift

Vals Legal Bench

14 rows · 0.2% avg lift

Missing frontier models

grok-4-1-fast-reasoning

Thin evidence after weighting

Rank #12

19.5%

grok-4-1-fast-non-reasoning

Thin evidence after weighting

Rank #15

14.9%

▶Taxonomy & task details

Core tasks

task.agentic_multi_step_completiontask.code_generation

Required modes

mode.tool_calling

Domains

domain.software_engineering

Related in Developer

Codebase onboarding brief

Summarize a repository's architecture, modules, and conventions.

Code generation

Generate correct, secure code from requirements.

Refactoring assistant

Refactor code safely while preserving behavior and improving clarity.

IDE code completion

Fast local-context code completion and small snippet generation.