FrontierSWE-class long-task coding scores across models.
Long-horizon coding benchmarks
Takeaway: GLM-5.2 is within ~1 point of Opus 4.8 and ahead of GPT-5.5.
Long-horizon coding
FrontierSWE, PostTrainBench, SWE-Marathon: GLM-5.2 vs the frontier.
Takeaway: GLM-5.2 is within ~1 point of Opus 4.8 and ahead of GPT-5.5.
Long-horizon coding
GLM-5.2 is within ~1 point of Opus 4.8 and ahead of GPT-5.5.
Rendering the interactive chart…
Expand data table
Data table
Download CSV| Glm51 | Glm52 | Gpt55 | Gemini | Opus48 | Series | Benchmark | Source | Value basis |
|---|---|---|---|---|---|---|---|---|
| 30.5 | 74.4 | 72.6 | 39.6 | 75.1 | coding_long | FrontierSWE | Z.ai: GLM-5.2: Built for Long-Horizon Tasks (2026-06-16) | Actual |
| 20.1 | 34.3 | 28.4 | 21.6 | 37.2 | coding_long | PostTrainBench | Z.ai: GLM-5.2: Built for Long-Horizon Tasks (2026-06-16) | Actual |
| 1 | 13 | 12 | 4 | 26 | coding_long | SWE-Marathon | Z.ai: GLM-5.2: Built for Long-Horizon Tasks (2026-06-16) | Actual |
How this evidence was built
What it measures
Source-backed values for all six charts come from the Z.ai glm-5.2 launch post (2026-06-16): long-horizon coding (FrontierSWE, PostTrainBench, swe-Marathon), standard coding (Terminal-Bench 2.1, swe-bench Pro, NL2Repo, DeepSWE, ProgramBench), reasoning (hle, CritPt, aime, hmmt, gpqa-Diamond, IMOAnswerBench), agentic (mcp-Atlas, Tool-Decathlon), the glm-5.2 vs glm-5.1 generational leap, and the mtp acceptance-length ablation.
Key context
- Every numeric point carries a source citation and a basis label naming the table row.
- Scores are reproduced verbatim from the Full Benchmark Table; cells the vendor published as "-" (not reported) are omitted, never estimated.
Key numbers
- FrontierSWE: glm-5.2 74.4 vs Opus 4.8 75.1 vs gpt-5.5 72.6 vs Gemini 39.6 vs glm-5.1 30.5 (zai-glm52).
- Terminal-Bench 2.1: glm-5.2 81 vs Opus 4.8 85 vs gpt-5.5 84 vs Gemini 74 (zai-glm52).
- Generational leap: swe-Marathon 13.0 (5.2) vs 1.0 (5.1) (zai-glm52).
- Mtp ablation quantifies the multi-token-prediction contribution to accepted length (zai-glm52).
Caveats
Caveat: these are the model developer’s self-reported figures under their stated harnesses and prompts (see the post’s footnotes); some competitor hle/CritPt cells are full-set scores. They are comparison-as-published, not an independent eval. Architecture context (not charted): 1M-token context (up from 200K), IndexShare cuts per-token indexer FLOPs ~2.9× at 1M, mit-licensed open weights.
Caveat: Some competitor scores are marked full-set vs public-set, not always like-for-like.
Do not use this to
- Independent verification: every figure is vendor self-reported (zai-glm52).
- Real-world reliability, latency or cost (benchmark scores only).
Scope
June 2026
Sources
Last verified 10 Sept 2026