GLM-5
Z.AIOpen weightsNon-reasoningVerifiedReleased Feb 12, 2026Input: text
Overall score
54.9
Rank 70 of 222 · 44.2–65.5
Context window
200K
200,000 tokens
Price per 1M tokens
$1.00 in · $3.20 out
Source: BenchLM
Blended price
$1.55
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Multilingual#689.5
- 2Instruction following#3588.6
- 3Math83.0
- 4Reasoning70.1
- 5Knowledge#8260.1
- 6Coding#7154.3
- 7Agentic & tool use42.9
Benchmark results (37)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| Gert Labs | 50.99 | #20 of 52 |
| BrowseComp | 62.0 | #38 of 50 |
| MCP Atlas | 31.1 | #41 of 42 |
| Terminal-Bench 2.0 | 56.2 | #24 of 39 |
| Claw-Eval | 57.7 | #27 of 36 |
| HLE w/ tools | 50.4 | #21 of 27 |
| CyberGym | 43.2 | #25 of 25 |
| τ³-bench results | 65.6 | #13 of 23 |
| Toolathlon | 38.0 | #18 of 22 |
| WideResearch | 69.8 | #13 of 16 |
| QwenClawBench | 54.1 | #6 of 10 |
| DeepPlanning | 14.6 | #6 of 7 |
| MCP-Tasks | 60.8 | #4 of 5 |
Coding
| SWE-bench Verified | 77.8 | #27 of 98 |
| SWE-bench Pro | 55.1 | #48 of 69 |
| NL2Repo | 35.9 | #27 of 31 |
| React Native Evals | 74.8 | #10 of 15 |
| SWE-Rebench | 62.8 | #2 of 13 |
| SWE-bench Verified* | 72.8 | #3 of 5 |
Math
| AIME26 | 95.8 | #7 of 29 |
| HMMT Feb 2026 | 86.4 | #13 of 24 |
| HMMT Feb 2025 | 97.5 | #1 of 13 |
| HMMT Nov 2025 | 96.9 | #1 of 12 |
| MMAnswerBench | 82.5 | #6 of 10 |
| AIME25 (Arcee) | 93.3 | #4 of 5 |
Knowledge
| GPQA | 86.0 | #44 of 117 |
| HLE | 30.5 | #43 of 69 |
| MMLU-Pro | 85.7 | #16 of 67 |
| SuperGPQA | 66.8 | #12 of 20 |
| MMLU-Pro (Arcee) | 85.8 | #3 of 5 |
Multilingual
| SWE Multilingual | 73.3 | #28 of 46 |
| MMLU-ProX | 83.1 | #6 of 16 |
| NOVA-63 | 55.1 | #7 of 7 |
Reasoning
| LongBench v2 | 60.8 | #10 of 17 |
| AI-Needle | 63.3 | #4 of 4 |
Multimodal
| Design Arena Website | 1247 | #47 of 91 |
Instruction following
| IFEval | 92.6 | #11 of 53 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price (BenchLM) | $1.00 | $3.20 | — |
| OpenRouter z-ai/glm-5 | $1.00 | $3.20 | $0.20 |
Z.AI's official pricing page lists GLM-5 at $1.00 input / $3.20 output per million tokens.
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text | 1446 | #60 · 29.3K votes |
| WebDev (coding) | 1435 | #83 · 7.5K votes |
Open weights
From the Hugging Face repository.
- Parameters
- 754B
- Licence
- mit
- Downloads (30 days)
- 907.9K
- Repository
- zai-org/GLM-5 ↗
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Similar score
- MiniMax M355.0 · $0.53
- Qwen 3.6 Max (preview)55.1 · —
- Agents-A154.4 · —
- Agents-A1-F16-GGUF54.4 · —
- Agents-A1-FP854.4 · —