Model comparison
GLM-5.1 vs Qwen3.5 397B
Head-to-head evidence from 28 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GLM-5.1 #18 (Supported); Qwen3.5 397B #71 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GLM-5.1 and Qwen3.5 397B share 28 comparable benchmark results. 4 of 8 categories are comparable. 8 results are unique to GLM-5.1; 27 to Qwen3.5 397B.
Updated July 23, 2026- Shared results
- 28
- GLM-5.1 only
- 8
- Qwen3.5 397B only
- 27
- Comparable categories
- 4 / 8
Pick GLM-5.1 if you want the stronger benchmark profile. Qwen3.5 397B only becomes the better choice if mathematics is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 28 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-5.1 is clearly ahead on the BenchAlign aggregate, 67.74 to 57.01. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
GLM-5.1's sharpest advantage is in agentic, where it averages 65.4 against 56.5. The single biggest benchmark swing on the page is HLE, 52.3% to 28.7%. Qwen3.5 397B does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.
GLM-5.1 is also the more expensive model on tokens at $1.40 input / $4.40 output per 1M tokens, versus $0.60 input / $3.60 output per 1M tokens for Qwen3.5 397B. GLM-5.1 is the reasoning model in the pair, while Qwen3.5 397B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. GLM-5.1 gives you the larger context window at 203K, compared with 128K for Qwen3.5 397B.
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-5.1 | Δ | Qwen3.5 397B |
|---|---|---|---|
| Math | GLM-5.162.0 | Margin→ 28.6 | Qwen3.5 397B90.6 |
| Agentic | GLM-5.165.4 | Margin← 8.9 | Qwen3.5 397B56.5 |
| Coding | GLM-5.161.3 | Margin→ 5.2 | Qwen3.5 397B66.5 |
| Knowledge | GLM-5.152.3 | Margin→ 4.3 | Qwen3.5 397B56.6 |
| Reasoning | GLM-5.1Not measured | MarginNo overlap | Qwen3.5 397B63.2 |
| Multilingual | GLM-5.1Not measured | MarginNo overlap | Qwen3.5 397B84.7 |
| Multimodal | GLM-5.1Not measured | MarginNo overlap | Qwen3.5 397B79.6 |
| Inst. Following | GLM-5.1Not measured | MarginNo overlap | Qwen3.5 397B92.6 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
HLE
KnowledgeA 52.3%B 28.7%Winner: GLM-5.1Δ 23.6HLE: GLM-5.1 scored 52.3%; Qwen3.5 397B scored 28.7%. GLM-5.1 wins this benchmark. - Source ↗
Terminal-Bench 2.0
AgenticA 63.5%B 52.5%Winner: GLM-5.1Δ 11Terminal-Bench 2.0: GLM-5.1 scored 63.5%; Qwen3.5 397B scored 52.5%. GLM-5.1 wins this benchmark. - Source ↗
SWE-bench Pro
CodingA 58.4%B 50.9%Winner: GLM-5.1Δ 7.5SWE-bench Pro: GLM-5.1 scored 58.4%; Qwen3.5 397B scored 50.9%. GLM-5.1 wins this benchmark. - Source ↗
BrowseComp
AgenticA 68%B 62%Winner: GLM-5.1Δ 6BrowseComp: GLM-5.1 scored 68%; Qwen3.5 397B scored 62%. GLM-5.1 wins this benchmark. - Source ↗
HMMT Feb 2026
MathA 82.6%B 87.9%Winner: Qwen3.5 397BΔ 5.3HMMT Feb 2026: GLM-5.1 scored 82.6%; Qwen3.5 397B scored 87.9%. Qwen3.5 397B wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GLM-5.1 | Qwen3.5 397B | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GLM-5.1$1.4 input / $4.4 output | Qwen3.5 397B$0.6 input / $3.6 output | Qwen3.5 397B has the lower combined listed price. |
| Generation speedtokens per second | GLM-5.1Not available | Qwen3.5 397B96 tok/s | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | GLM-5.1Not available | Qwen3.5 397B2.44 s | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | GLM-5.1203K | Qwen3.5 397B128K | GLM-5.1 lists the larger context window. |
Benchmark Deep Dive
AgenticGLM-5.1 wins19 benchmarks
| Benchmark | GLM-5.1 | Qwen3.5 397B | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 63.5% | 52.5% | GLM-5.1 leads |
| BrowseCompSource | 68% | 62% | GLM-5.1 leads |
| τ³-bench resultsSource | 70.6% | 68.4% | GLM-5.1 leads |
| MCP AtlasSource | 71.8% | 46.1% | GLM-5.1 leads |
| CyberGymSource | 68.7% | — | Not comparable |
| Claw-EvalSource | 62.3% | 56.8% | GLM-5.1 leads |
| AA Agentic IndexSource | 29.9% | 19.9% | GLM-5.1 leads |
| τ²-bench resultsSource | 97.7% | 95.6% | GLM-5.1 leads |
| GDPval-AASource | 37.8% | 23.1% | GLM-5.1 leads |
| Gert LabsSource | 60.11% | 46.76% | GLM-5.1 leads |
| GDPval-AASource | 1257 | 962 | GLM-5.1 leads |
| ResearchClawBenchSource | 18.2% | 14.2% | GLM-5.1 leads |
| QwenClawBenchSource | — | 51.8% | Not comparable |
| VITA-BenchSource | — | 43.7% | Not comparable |
| DeepPlanningSource | — | 37.6% | Not comparable |
| ToolathlonSource | — | 36.3% | Not comparable |
| MCP-TasksSource | — | 74.2% | Not comparable |
| WideResearchSource | — | 74.0% | Not comparable |
| APEX-Agents-AASource | — | 15.3% | Not comparable |
CodingQwen3.5 397B wins8 benchmarks
| Benchmark | GLM-5.1 | Qwen3.5 397B | Result |
|---|---|---|---|
| SWE-bench ProSource | 58.4% | 50.9% | GLM-5.1 leads |
| NL2RepoSource | 42.7% | — | Not comparable |
| SWE-RebenchSource | 62.7% | — | Not comparable |
| Vibe Code BenchSource | 31.46% | — | Not comparable |
| AA Coding IndexSource | 55.8% | 48.2% | GLM-5.1 leads |
| AA-SciCodeSource | 43.8% | 42.0% | GLM-5.1 leads |
| SWE-bench VerifiedSource | — | 76.2% | Not comparable |
| LiveCodeBench v6Source | — | 83.6% | Not comparable |
Reasoning4 benchmarks
KnowledgeQwen3.5 397B wins13 benchmarks
| Benchmark | GLM-5.1 | Qwen3.5 397B | Result |
|---|---|---|---|
| GPQA-DSource | 86.2% | — | Not comparable |
| HLESource | 52.3% | 28.7% | GLM-5.1 leads |
| Artificial Analysis Intelligence IndexSource | 40.2% | 33.7% | GLM-5.1 leads |
| AA-GPQA DiamondSource | 86.8% | 89.3% | Qwen3.5 397B leads |
| AA-HLESource | 28.0% | 27.3% | GLM-5.1 leads |
| AA-Omniscience IndexSource | 1.9% | -29.8% | GLM-5.1 leads |
| AA-Omniscience AccuracySource | 24.2% | 31.4% | Qwen3.5 397B leads |
| AA-Omniscience Hallucination RateSource | 29.4% | 89.1% | GLM-5.1 leads |
| GPQASource | — | 88.4% | Not comparable |
| SuperGPQASource | — | 70.4% | Not comparable |
| MMLU-ProSource | — | 87.8% | Not comparable |
| MMLU-ReduxSource | — | 94.9% | Not comparable |
| C-EvalSource | — | 93% | Not comparable |
MathQwen3.5 397B wins7 benchmarks
| Benchmark | GLM-5.1 | Qwen3.5 397B | Result |
|---|---|---|---|
| AIME26Source | 95.3% | 93.3% | GLM-5.1 leads |
| HMMT Nov 2025Source | 94.0% | 92.7% | GLM-5.1 leads |
| HMMT Feb 2026Source | 82.6% | 87.9% | Qwen3.5 397B leads |
| MMAnswerBenchSource | 83.8% | 80.9% | GLM-5.1 leads |
| FrontierMath v2 (Tiers 1-3)Source | 33.448% | — | Not comparable |
| FrontierMath v2 (Tier 4)Source | 12.500% | — | Not comparable |
| HMMT Feb 2025Source | — | 94.8% | Not comparable |
Multilingual2 benchmarks
Multimodal8 benchmarks
| Benchmark | GLM-5.1 | Qwen3.5 397B | Result |
|---|---|---|---|
| Design Arena WebsiteSource | 1305 | — | Not comparable |
| MMMU-ProSource | — | 79% | Not comparable |
| MathVisionSource | — | 88.6% | Not comparable |
| CharXivSource | — | 80.8% | Not comparable |
| VideoMMMUSource | — | 84.7% | Not comparable |
| ScreenSpot ProSource | — | 65.6% | Not comparable |
| V*Source | — | 95.8% | Not comparable |
| AA-MMMU-ProSource | — | 77.3% | Not comparable |
Frequently Asked Questions (5)
Which is better, GLM-5.1 or Qwen3.5 397B?
GLM-5.1 is ahead on BenchLM's BenchAlign leaderboard, 67.74 to 57.01. The biggest single separator in this matchup is HLE, where the scores are 52.3% and 28.7%.
Which is better for knowledge tasks, GLM-5.1 or Qwen3.5 397B?
Qwen3.5 397B has the edge for knowledge tasks in this comparison, averaging 56.6 versus 52.3. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.
Which is better for coding, GLM-5.1 or Qwen3.5 397B?
Qwen3.5 397B has the edge for coding in this comparison, averaging 66.5 versus 61.3. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.
Which is better for math, GLM-5.1 or Qwen3.5 397B?
Qwen3.5 397B has the edge for math in this comparison, averaging 90.6 versus 62. Inside this category, HMMT Feb 2026 is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GLM-5.1 or Qwen3.5 397B?
GLM-5.1 has the edge for agentic tasks in this comparison, averaging 65.4 versus 56.5. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
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.