Model comparison
GLM-4.7 vs GLM-5.2
Head-to-head evidence from 19 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GLM-4.7 #42 (Supported); GLM-5.2 #37 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GLM-4.7 and GLM-5.2 share 19 comparable benchmark results. 4 of 8 categories are comparable. 11 results are unique to GLM-4.7; 24 to GLM-5.2.
Updated July 23, 2026- Shared results
- 19
- GLM-4.7 only
- 11
- GLM-5.2 only
- 24
- Comparable categories
- 4 / 8
Pick GLM-5.2 if you want the stronger benchmark profile. GLM-4.7 only becomes the better choice if coding is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 19 shared benchmark results across 6 evidence categories; 4 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GLM-5.2 has the cleaner BenchAlign overall profile here, landing at 63.96 versus 61.16. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
GLM-5.2's sharpest advantage is in mathematics, where it averages 95.9 against 1.8. The single biggest benchmark swing on the page is Terminal-Bench 2.0, 41% to 81%. GLM-4.7 does hit back in coding, so the answer changes if that is the part of the workload you care about most.
GLM-5.2 is also the more expensive model on tokens at $1.40 input / $4.40 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for GLM-4.7. That is roughly Infinityx on output cost alone. GLM-5.2 gives you the larger context window at 1M, compared with 200K for GLM-4.7.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | GLM-4.7 | Δ | GLM-5.2 |
|---|---|---|---|
| Math | GLM-4.71.8 | Margin→ 94.1 | GLM-5.295.9 |
| Agentic | GLM-4.745.7 | Margin→ 35.3 | GLM-5.281.0 |
| Coding | GLM-4.775.4 | Margin← 13.3 | GLM-5.262.1 |
| Knowledge | GLM-4.751.8 | Margin→ 7.8 | GLM-5.259.6 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
Terminal-Bench 2.0
AgenticA 41%B 81%Winner: GLM-5.2Δ 40Terminal-Bench 2.0: GLM-4.7 scored 41%; GLM-5.2 scored 81%. GLM-5.2 wins this benchmark. - Source ↗
HLE
KnowledgeA 24.8%B 54.7%Winner: GLM-5.2Δ 29.9HLE: GLM-4.7 scored 24.8%; GLM-5.2 scored 54.7%. GLM-5.2 wins this benchmark. - Source ↗
GPQA
KnowledgeA 85.7%B 91.2%Winner: GLM-5.2Δ 5.5GPQA: GLM-4.7 scored 85.7%; GLM-5.2 scored 91.2%. GLM-5.2 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GLM-4.7 | GLM-5.2 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GLM-4.7$0 input / $0 output | GLM-5.2$1.4 input / $4.4 output | GLM-4.7 has the lower combined listed price. |
| Generation speedtokens per second | GLM-4.782 tok/s | GLM-5.2Not available | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | GLM-4.71.10 s | GLM-5.2Not available | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | GLM-4.7200K | GLM-5.21M | GLM-5.2 lists the larger context window. |
Benchmark Deep Dive
AgenticGLM-5.2 wins20 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 41% | 81% | GLM-5.2 leads |
| BrowseCompSource | 52% | — | Not comparable |
| VITA-BenchSource | 15.5% | — | Not comparable |
| AA Agentic IndexSource | 25.4% | 43.1% | GLM-5.2 leads |
| τ²-bench resultsSource | 95.9% | 99.1% | GLM-5.2 leads |
| Gert LabsSource | 39.95% | — | Not comparable |
| GDPval-AASource | 33.3% | 50.7% | GLM-5.2 leads |
| GDPval-AASource | 1165 | 1514 | GLM-5.2 leads |
| MCP AtlasSource | — | 76.8% | Not comparable |
| ToolathlonSource | — | 48.2% | Not comparable |
| APEX-Agents-AASource | — | 33.7% | Not comparable |
| ResearchClawBenchSource | — | 20.7% | Not comparable |
| AA BriefcaseSource | — | 1260 | Not comparable |
| AA AutomationBenchSource | — | 27.8% | Not comparable |
| AA EnterpriseOps-GymSource | — | 42.7% | Not comparable |
| AA Harvey LABSource | — | 91.0% | Not comparable |
| AA ITBenchSource | — | 42.7% | Not comparable |
| AA Tau3 BankingSource | — | 26.8% | Not comparable |
| terminalBenchHardSource | — | 50.8% | Not comparable |
| aaTerminalBench21Source | — | 77.9% | Not comparable |
CodingGLM-4.7 wins11 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | 73.8% | — | Not comparable |
| LiveCodeBenchSource | 84.9% | — | Not comparable |
| SWE-RebenchSource | 58.7% | — | Not comparable |
| AA Coding IndexSource | 45.3% | 68.8% | GLM-5.2 leads |
| AA-SciCodeSource | 45.1% | 50.5% | GLM-5.2 leads |
| AA LiveCodeBenchSource | 89.4% | — | Not comparable |
| SWE-bench ProSource | — | 62.1% | Not comparable |
| NL2RepoSource | — | 48.9% | Not comparable |
| Terminal-Bench 2.0Source | — | 81.0% | Not comparable |
| ProgramBenchSource | — | 63.7% | Not comparable |
| cursorBench32Source | — | 55.0% | Not comparable |
Reasoning2 benchmarks
KnowledgeGLM-5.2 wins12 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| GPQASource | 85.7% | 91.2% | GLM-5.2 leads |
| MMLU-ProSource | 84.3% | — | Not comparable |
| HLESource | 24.8% | 54.7% | GLM-5.2 leads |
| Artificial Analysis Intelligence IndexSource | 33.7% | 51.1% | GLM-5.2 leads |
| AA-GPQA DiamondSource | 85.9% | 89.5% | GLM-5.2 leads |
| AA-HLESource | 25.1% | 40.1% | GLM-5.2 leads |
| AA-Omniscience IndexSource | -34.6% | 4.0% | GLM-5.2 leads |
| AA-Omniscience AccuracySource | 29.3% | 25.1% | GLM-4.7 leads |
| AA-Omniscience Hallucination RateSource | 90.3% | 28.1% | GLM-5.2 leads |
| GPQA-DSource | — | 91.2% | Not comparable |
| HLE w/o toolsSource | — | 40.5% | Not comparable |
| AA Openness IndexSource | — | 44.4% | Not comparable |
MathGLM-5.2 wins7 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| AIME 2025Source | 95.7% | — | Not comparable |
| FrontierMath v2 (Tiers 1-3)Source | 2.439% | — | Not comparable |
| FrontierMath v2 (Tier 4)Source | 0.000% | — | Not comparable |
| AIME26Source | — | 99.2% | Not comparable |
| HMMT Nov 2025Source | — | 94.4% | Not comparable |
| HMMT Feb 2026Source | — | 92.5% | Not comparable |
| MMAnswerBenchSource | — | 91.0% | Not comparable |
Multimodal1 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| Design Arena WebsiteSource | 1255 | 1340 | GLM-5.2 leads |
Inst. Following1 benchmarks
| Benchmark | GLM-4.7 | GLM-5.2 | Result |
|---|---|---|---|
| AA-IFBenchSource | 67.9% | 73.3% | GLM-5.2 leads |
Frequently Asked Questions (5)
Which is better, GLM-4.7 or GLM-5.2?
GLM-5.2 is ahead on BenchLM's BenchAlign leaderboard, 63.96 to 61.16. The biggest single separator in this matchup is Terminal-Bench 2.0, where the scores are 41% and 81%.
Which is better for knowledge tasks, GLM-4.7 or GLM-5.2?
GLM-5.2 has the edge for knowledge tasks in this comparison, averaging 59.6 versus 51.8. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.
Which is better for coding, GLM-4.7 or GLM-5.2?
GLM-4.7 has the edge for coding in this comparison, averaging 75.4 versus 62.1. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.
Which is better for math, GLM-4.7 or GLM-5.2?
GLM-5.2 has the edge for math in this comparison, averaging 95.9 versus 1.8. GLM-4.7 stays close enough that the answer can still flip depending on your workload.
Which is better for agentic tasks, GLM-4.7 or GLM-5.2?
GLM-5.2 has the edge for agentic tasks in this comparison, averaging 81 versus 45.7. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.