Compare / head-to-head
GLM-5.3vs
Qwen3.5 2B
GLM-5.3 leads 1 of 1 shared benchmark. GLM-5.3 has the larger context window (1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
1 – 0
GLM-5.3 leads
Cheaper per token
—
list price, input + output
Larger context
GLM-5.3
1M tokens
Providers
2 providers
Z.ai · Qwen
GLM-5.3
Z.ai · released 2026-08-18
text
Qwen3.5 2B
Qwen · released 2026-03-02
textvisionvideo
- Context
- 262K
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 16 · 5 core
Quality
Benchmark matrix
1 shared · 42 only GLM-5.3 · 15 only Qwen3.5 2B| Benchmark | GLM-5.3 | Qwen3.5 2B | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 1 | ||||
| Chess Puzzles (Epoch AI run) | 21% ↗epoch run | 0% ↗epoch run | +21 pt | GLM-5.3 |
| Only GLM-5.3 reports · 42 | ||||
| GPQA Diamond | 90.9% ↗epoch run | not reported | — | — |
| LMArena Elo | 1478.5 ↗ | not reported | — | — |
| Agents' Last Exam | 28.5% ↗ | not reported | — | — |
| Agents' Last Exam (ALE-CLI) | 28.5% ↗ | not reported | — | — |
| APEX-Agents (Mercor) | 56.6% ↗ | not reported | — | — |
| AutomationBench v1.0.6 | 48.2% ↗ | not reported | — | — |
| CyberGym | 84.5% ↗ | not reported | — | — |
| DeepSWE v1.1 | 66.9% ↗ | not reported | — | — |
| DeepSWE v1.1 (Datacurve) | 69% ↗ | not reported | — | — |
| ExploitBench | 54.4% ↗ | not reported | — | — |
| ExploitGym 2h | 105 ↗ | not reported | — | — |
| ExploitGym 6h | 130 ↗ | not reported | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | 29.3% ↗epoch run | not reported | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 68.8% ↗epoch run | not reported | — | — |
| FrontierSWE | 78.1% ↗ | not reported | — | — |
| FrontierSWE V2 (Proximal Labs) | 30.2% ↗ | not reported | — | — |
| GDPval-AA v2 | 1769 ↗ | not reported | — | — |
| HLE (with tools) | 62.5% ↗ | not reported | — | — |
| LiveBench Agentic Coding (LiveBench) | 60.9% ↗ | not reported | — | — |
| LiveBench Coding (LiveBench) | 79% ↗ | not reported | — | — |
| LiveBench Data Analysis (LiveBench) | 70.2% ↗ | not reported | — | — |
| LiveBench Instruction Following (LiveBench) | 69.3% ↗ | not reported | — | — |
| LiveBench Language (LiveBench) | 79.9% ↗ | not reported | — | — |
| LiveBench Mathematics (LiveBench) | 87.9% ↗ | not reported | — | — |
| LiveBench Reasoning (LiveBench) | 85.8% ↗ | not reported | — | — |
| LMArena Agent (LMArena) | 0.0236 ↗ | not reported | — | — |
| LMArena WebDev (LMArena) | 1623 ↗ | not reported | — | — |
| Mystery Game Puzzles (Epoch AI run) | 33% ↗epoch run | not reported | — | — |
| NL2Repo | 58% ↗ | not reported | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 91.1% ↗epoch run | not reported | — | — |
| PostTrainBench | 39.8% ↗ | not reported | — | — |
| ProgramBench (Almost Solved) | 19% ↗ | not reported | — | — |
| SimpleBench (SimpleBench) | 66.2% ↗ | not reported | — | — |
| SimpleQA Verified | 41% ↗epoch run | not reported | — | — |
| SWE-Marathon v1.1 | 42.5% ↗ | not reported | — | — |
| Terminal-Bench 2.1 | 88.2% ↗ | not reported | — | — |
| Terminal-Bench 3.0 | 28.3% ↗ | not reported | — | — |
| Terminal-Bench 4.0 (Vals AI) | 38.9% ↗ | not reported | — | — |
| Toolathlon Verified | 73% ↗ | not reported | — | — |
| Vending-Bench 2 (Andon Labs) | 8163.61 ↗ | not reported | — | — |
| WeirdML (Håvard Tveit Ihle) | 75.4% ↗ | not reported | — | — |
| Z.ai Code Bench (max effort) | 34.5% ↗ | not reported | — | — |
| Only Qwen3.5 2B reports · 15 | ||||
| BFCL-V4 (thinking) | not reported | 43.6% ↗ | — | — |
| GPQA (thinking) | not reported | 51.6% ↗ | — | — |
| HMMT Feb 25 (thinking) | not reported | 22.9% ↗ | — | — |
| IFBench (thinking) | not reported | 41.3% ↗ | — | — |
| IFEval (non-thinking) | not reported | 61.2% ↗ | — | — |
| IFEval (thinking) | not reported | 78.6% ↗ | — | — |
| LongBench v2 (thinking) | not reported | 38.7% ↗ | — | — |
| MMLU-Pro (non-thinking) | not reported | 55.3% ↗ | — | — |
| MMLU-Pro (thinking) | not reported | 66.5% ↗ | — | — |
| MMLU-Redux (thinking) | not reported | 79.6% ↗ | — | — |
| MMMLU (thinking) | not reported | 63.1% ↗ | — | — |
| MMMU (thinking) | not reported | 64.2% ↗ | — | — |
| MMMU-Pro (thinking) | not reported | 50.3% ↗ | — | — |
| SuperGPQA (thinking) | not reported | 37.5% ↗ | — | — |
| TAU2-Bench (thinking) | not reported | 48.8% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GLM-5.3 minus Qwen3.5 2B in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GLM-5.3 | Qwen3.5 2B | Edge |
|---|---|---|---|
| Context window | 1M tokens | 262K tokens | GLM-5.3 |
| Max output | 131K tokens | — | — |
| Input price / 1M | $1.4 ↗ | — | — |
| Output price / 1M | $4.4 ↗ | — | — |
| Cached input / 1M | $0.26 ↗ | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $5.8 | — | — |
| Modalities | text | text · vision · video | Qwen3.5 2B |
| Released | 2026-08-18 | 2026-03-02 | — |
| Cited benchmark scores | 57 | 16 | — |
Reliability
Provider status
Live from /statusMore matchups
GLM-5.3 vs …
Models sharing the most benchmarksMore matchups
Qwen3.5 2B vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GLM-5.3 and Qwen3.5 2B on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.