GLM-5.1
Language modelOpen source
GLM-5.1
Z.ai
- Released
- April 7, 2026
- Data date
- September 8, 2026
- Predecessor
- GLM-5
- Speed
- 56.3 tokens/sOutput throughput
- Price
- Input · Output / 1M tokens
- Input
- Text
- Output
- Text
GLM-5.1 keeps the 744B base of GLM-5 and makes stability on long tasks its documented focus. It activates 40B parameters per token and provides a 200K context. The openly available MoE model remains aimed at demanding text and coding work.
Z.ai documents reasoning and direct responses as separate operating modes. The large parameter count describes the full MoE base, not the compute used for every token. The 744B figure therefore does not mean that all parameters run for every token.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | Open weightsSource |
| Access | API, LocalSource |
| Parameters | 744BSource |
| Active parameters | 40BSource |
| Context window | 200,000 tokensSource |
| Notes | MIT-licensed model with 744B total parameters, 40B active, and a 200K context window.Source |
| License | MITSource |
| Architecture | MoE with DeepSeek Sparse AttentionSource |
| Parameters | 744 BSource |
| Active parameters | 40 BSource |
| Input | TextSource |
| Output | TextSource |
| Reasoning mode | thinking, non-thinkingSource |
| Tool calling | YesSource |
| API model ID | glm-5.1Source |
Page 1 of 16
Published benchmarks
LMArena Text, Style Control
Russian
Note (GLM-5.1): The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.
1.GPT-5.51,478.72 rating
Note: The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.
higher is betterAs of October 15, 2026Scale: 0 rating - 1,478.72 rating
Methodology and context
- Comparison cohort
- lmarena-text-style-control:2026-10-13:russian
- Metric
- style-controlled Bradley-Terry rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 2
- Methodology and context
- Arena rating from 7,421 published pairwise votes.. Source model ID: lmarena/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Z.ai. Model snapshot: glm-5.1. Evaluator: LMArena. Dataset type: public-crowdsourced-preferences. Leaderboard date: 2026-10-13. Harness: Arena Text, style control. Source ID: lmarena. Evidence capture ID: lmarena-4589dee83b594891961ae5190c508a9288d2c2ae-a39cbed99996. Source payload hash: a39cbed99996f4993a4cd66d8322b2edb939f1ffdbd377c9d3769edd01dd78ca. Parser version: lmarena-dataset-v1. Imported at: 2026-10-15
Coding
Code Migration
| Sub-task | GLM-5.1 (This model) | Nemotron 3.5 Lightning |
|---|---|---|
| CLI | 18.83% (best value in this row) | 3.94% |
| COBOL | 46.57% (best value in this row) | 0% |
| Overall | 25.77% (best value in this row) | 2.95% |
higher is betterAs of October 15, 2026Source: Vals AIbars scaled within each row, not across rows
Methodology and context
- Metric
- accuracy
Code Migration
Code Quality
higher is betterAs of October 15, 2026Scale: 0 rating - 3.54 rating
Methodology and context
- Comparison cohort
- vals-code-migration:1:code_quality
- Metric
- mean reviewer code-quality rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 2
- Methodology and context
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-12. Temperature: 1. Top P: 0.95. Maximum output tokens: 131072. Latency in seconds: 3502.363. Standard error: 0.061. Cost per task in USD: 5.701313. Score display: 3.544 / 5. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-10-15-3fa9c10568ec. Source payload hash: 3fa9c10568ecc117fabec81767c405b49a25d5d2ded308ee435fffb197c53474. Parser version: vals-astro-v3. Imported at: 2026-10-15
Finance
CorpFin v2
| Sub-task | GLM-5.1 (This model) | Claude Opus 5 | GLM-5.2 |
|---|---|---|---|
| Exact Pages | 61.77% | 68.65% (best value in this row) | 63.64% |
| Max Fitting Context | 66.32% | 75.99% (best value in this row) | 68.42% |
| Overall | 64.45% | 73.19% (best value in this row) | 66.12% |
| Shared Max Context | 65.27% | 74.94% (best value in this row) | 66.32% |
higher is betterAs of October 15, 2026Source: Vals AIbars scaled within each row, not across rows
Methodology and context
- Metric
- accuracy
Recorded measurement details
LMArena Text, Style Control: Russian (Show measurement, test conditions, and source)
- Source value
- 1457.7955225178212
- Score
- 1457.7955225178212
- Metric
- style-controlled Bradley-Terry rating
- Unit
- rating
- Benchmark version
- style-control-v1
- Category
- general
- Task
- russian
- Task label
- Russian
- Direction
- Higher is better
- Comparison cohort
- lmarena-text-style-control:2026-10-13:russian
- Confidence interval
- 7.444491
- Source type
- official-leaderboard
- Evaluator
- LMArena
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Arena rating from 7,421 published pairwise votes.. Source model ID: lmarena/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Z.ai. Model snapshot: glm-5.1. Evaluator: LMArena. Dataset type: public-crowdsourced-preferences. Leaderboard date: 2026-10-13. Harness: Arena Text, style control. Source ID: lmarena. Evidence capture ID: lmarena-4589dee83b594891961ae5190c508a9288d2c2ae-a39cbed99996. Source payload hash: a39cbed99996f4993a4cd66d8322b2edb939f1ffdbd377c9d3769edd01dd78ca. Parser version: lmarena-dataset-v1. Imported at: 2026-10-15
- Comparison methodology
- Arena rating from 7,421 published pairwise votes.. Harness: Arena Text, style control
- Context
- The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.
Code Migration: CLI (Show measurement, test conditions, and source)
- Source value
- 18.833
- Score
- 18.833
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- cli
- Task label
- CLI
- Direction
- Higher is better
- Comparison cohort
- vals-code-migration:1:cli
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-12. Temperature: 1. Top P: 0.95. Maximum output tokens: 131072. Latency in seconds: 3837.721. Standard error: 1.569. Cost per task in USD: 6.652074. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-10-15-3fa9c10568ec. Source payload hash: 3fa9c10568ecc117fabec81767c405b49a25d5d2ded308ee435fffb197c53474. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Temperature: 1. Top P: 0.95. Maximum output tokens: 131072
Code Migration: COBOL (Show measurement, test conditions, and source)
- Source value
- 46.571
- Score
- 46.571
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- cobol
- Task label
- COBOL
- Direction
- Higher is better
- Comparison cohort
- vals-code-migration:1:cobol
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-12. Temperature: 1. Top P: 0.95. Maximum output tokens: 131072. Latency in seconds: 2496.289. Standard error: 15.655. Cost per task in USD: 2.84903. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-10-15-3fa9c10568ec. Source payload hash: 3fa9c10568ecc117fabec81767c405b49a25d5d2ded308ee435fffb197c53474. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Temperature: 1. Top P: 0.95. Maximum output tokens: 131072
Code Migration: Overall (Show measurement, test conditions, and source)
- Source value
- 25.768
- Score
- 25.768
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-code-migration:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-12. Temperature: 1. Top P: 0.95. Maximum output tokens: 131072. Latency in seconds: 3502.363. Standard error: 4.087. Cost per task in USD: 5.701313. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-10-15-3fa9c10568ec. Source payload hash: 3fa9c10568ecc117fabec81767c405b49a25d5d2ded308ee435fffb197c53474. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Temperature: 1. Top P: 0.95. Maximum output tokens: 131072
Code Migration: Code Quality (Show measurement, test conditions, and source)
- Source value
- 3.544
- Score
- 3.544
- Metric
- mean reviewer code-quality rating
- Unit
- rating
- Benchmark version
- 1
- Category
- coding
- Task
- code_quality
- Task label
- Code Quality
- Direction
- Higher is better
- Comparison cohort
- vals-code-migration:1:code_quality
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-12. Temperature: 1. Top P: 0.95. Maximum output tokens: 131072. Latency in seconds: 3502.363. Standard error: 0.061. Cost per task in USD: 5.701313. Score display: 3.544 / 5. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-10-15-3fa9c10568ec. Source payload hash: 3fa9c10568ecc117fabec81767c405b49a25d5d2ded308ee435fffb197c53474. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Temperature: 1. Top P: 0.95. Maximum output tokens: 131072
CorpFin v2: Exact Pages (Show measurement, test conditions, and source)
- Source value
- 61.772
- Score
- 61.772
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- exact_pages
- Task label
- Exact Pages
- Direction
- Higher is better
- Comparison cohort
- vals-corp-fin-v2:2:exact_pages
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 18.86. Standard error: 1.659. Cost per task in USD: 0.005086. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-10-15-7cf2217ab372. Source payload hash: 7cf2217ab3728d5a84bffce82aab9fc52d7f810b0cbe8c65b387df47f3d31c9b. Parser version: vals-astro-v3. Imported at: 2026-10-15
CorpFin v2: Max Fitting Context (Show measurement, test conditions, and source)
- Source value
- 66.317
- Score
- 66.317
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- max_fitting_context
- Task label
- Max Fitting Context
- Direction
- Higher is better
- Comparison cohort
- vals-corp-fin-v2:2:max_fitting_context
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 139.915. Standard error: 1.614. Cost per task in USD: 0.15991. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-10-15-7cf2217ab372. Source payload hash: 7cf2217ab3728d5a84bffce82aab9fc52d7f810b0cbe8c65b387df47f3d31c9b. Parser version: vals-astro-v3. Imported at: 2026-10-15
CorpFin v2: Overall (Show measurement, test conditions, and source)
- Source value
- 64.452
- Score
- 64.452
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-corp-fin-v2:2:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 68.17. Standard error: 0.943. Cost per task in USD: 0.078922. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-10-15-7cf2217ab372. Source payload hash: 7cf2217ab3728d5a84bffce82aab9fc52d7f810b0cbe8c65b387df47f3d31c9b. Parser version: vals-astro-v3. Imported at: 2026-10-15
CorpFin v2: Shared Max Context (Show measurement, test conditions, and source)
- Source value
- 65.268
- Score
- 65.268
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- shared_max_context
- Task label
- Shared Max Context
- Direction
- Higher is better
- Comparison cohort
- vals-corp-fin-v2:2:shared_max_context
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: zai/glm-5.1. Source model name: GLM-5.1. Provider reported by source: Zhipu AI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 45.736. Standard error: 1.625. Cost per task in USD: 0.071771. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-10-15-7cf2217ab372. Source payload hash: 7cf2217ab3728d5a84bffce82aab9fc52d7f810b0cbe8c65b387df47f3d31c9b. Parser version: vals-astro-v3. Imported at: 2026-10-15
Page 1 of 16
Our tests
SVG short test
One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.
Earlier tests
Pelican on a bicycle
A community classic for free-form SVG drawing.
GLM-5.1
Output withheld for safety
Reasoning (requested): high
Current attempt
- Run date
- Oct 5, 2026
- API provider
- openrouter
- Requested API model
- z-ai/glm-5.1
- API model reported in response
- z-ai/glm-5.1
- Reasoning setting
- high
- Pinned endpoint
- Z.AI | z-ai/glm-5.1-20260406
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Failed
- Response time
- 141.6 s
Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.
Earlier attempts (3)
- Oct 2, 2026Output withheld for safetyReasoning setting: high
- Oct 1, 2026API request failedReasoning setting: high
- Sep 7, 2026API request failedReasoning setting: high
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)Popularity
43 / 100