DeepSWEmini-swe-agent
44%±2%
- bench
- DeepSWE
- version
- v1.1
- split
- public 113 tasks
- harness
- mini-swe-agent
- n
- 113
- metric
- Pass@1
- value
- 44%
- ci
- ±2%
- $/task
- $3.92
Open MIT weights at zai-org/GLM-5.2. Catalog $0.966 / $3.036, cache read $0.1932 /M, 1,048,576 / 131,072, created unix 1781631930 (16 Jun 2026).
5.3 is the current hosted id. This card will not put 5.3’s AA 60 or vals 71.48% on 5.2.
Use 5.2 when you want the open MIT checkpoint and the only Z.ai DeepSWE listing we fetched. Official mini-swe-agent: 44% ±2, $3.92/task. AA 53 at max. Hub API 19 Aug 2026: 5,011 likes, 2,748,563 downloads.
We did not open a vals.ai card for this identity. Chip omitted.
Ranked view: Best coding models: the independent leaderboard puts this row and every other card that carries an independent coding score on one board, with the confidence intervals left visible.
Copied from AA or vals HTML we opened. 0.0% placeholder bars are omitted. Do not average these into the hero tiles, and do not invent CI, $/task, or steps for Artificial Analysis.
Plus is $25/mo with $25 weekly hosted usage. The Mac app stays free with your own keys. Get Plus is not Download for Mac.
Community quants: Community quants are community.
This is the model repo, not Download for Mac. Downloads and likes from the Hugging Face API on 2026-08-19.
Not listed as a Continuum hosted id. Hosted GLM is glm-5.3. Run the MIT weights or OpenRouter z-ai/glm-5.2.
We list first-party and OpenRouter, plus Continuum only when the live public allowlist named the id. This is not a 15-host routing table.
DeepSWE official: $3.92 per task (44% ±2). That is the hero economics number.
Official Z.ai pricing.md (same table as 5.3), 19 Aug 2026: $1.4 / $0.26 cached / storage Limited-time Free / $4.4. OpenRouter 5.2 is $0.966 / $3.036, cache $0.1932. Cite both; do not collapse.
Artificial Analysis Intelligence Index 53 at max, fetched 19 Aug 2026. Same page: 103 tok/s; $843.44 Index eval. AA $1.40 / $4.40 matches official Z.ai and is above the OR pair.
| Claim | Source | As of |
|---|---|---|
| DeepSWE 44% ±2% at undefined | DeepSWE official board | 2026-08-13 |
| Official GLM-5.2 $1.4 / $0.26 cached / storage Limited-time Free / $4.4 (same table as 5.3) | Z.ai pricing.md | 2026-08-19 |
| OR catalog $0.966 / $3.036, cache $0.1932, 1,048,576 / 131,072, created 1781631930 | OpenRouter catalog | 2026-08-19 |
| DeepSWE 44% ±2, $3.92/task | DeepSWE official board | 2026-08-13 |
| AA Index 53 (max); $1.40 / $4.40; 103 tok/s; $843.44 Index eval | Artificial Analysis GLM-5.2 | 2026-08-19 |
| HF likes 5,011; downloads 2,748,563; usedStorage 1,506,689,458,421; safetensors.total 753,329,940,480; MIT; created 2026-06-16T07:39:20Z | Hugging Face API | 2026-08-19 |
| 4.34T week tokens (+18% WoW) | OpenRouter rankings This Week | 2026-08-19 |
| OpenRouter id z-ai/glm-5.2, context 1,048,576 | OpenRouter /api/v1/models | 2026-08-19 |
| HF downloads 2,748,563 | Hugging Face API zai-org/GLM-5.2 | 2026-08-19 |
Same official DeepSWE harness (mini-swe-agent, public 113 tasks), fetched 2026-08-19. Different effort labels are the lab's own setting on that board, shown here rather than normalized.
Continue through Z.ai's model family with GLM 5.1.
The 5.3 card already names GLM-5.2 at 44% ±2, $3.92/task on the official DeepSWE board. This chip is that 5.2 row, not a 5.3 integer pasted onto a different model. AA is 53 here and 60 on 5.3.
Lab posts pick a harness, an effort, a split, and sometimes a private eval set. DeepSWE publishes the official mini-swe-agent row with cost, tokens, and steps. Scale SWE-bench Pro only counts when the public shared-harness board has a row we can fetch. A higher lab number is usually a different test, not a better one.
No Continuum hosted id is verified for this model, so this page does not invent a curl target. Use the first-party API or the OpenRouter slug z-ai/glm-5.2.