Side by side
See the difference.
Scores, capabilities and costs for the models you’re considering.
Scroll to compare models →
| 2 models | googlegemma-3-4b-itRemove × | zai-orgGLM-5.3Remove × |
|---|---|---|
| What’s different | Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and… | Reuses the GLM-5.2 base with new post-training; exposes low, high and max reasoning effort, defaulting to max. |
| Date | Feb 20, 2025Listed | Aug 18, 2026Released |
API prices USD · provider configuration shown below | ||
| Listed rate | $0.04 in / $0.05 out/ 1M tokensParasail · via OpenRouter · all routes ↗parasail/bf16 · quantization=bf16 | $0.079 in / $1.40 out/ 1M tokensWafer · via OpenRouter · all routes ↗wafer/us · quantization=fp4 |
| Provider endpoints | 6 endpoints → | 59 endpoints → |
Benchmarks Each result retains its evaluation configuration | ||
| AA Intelligence ↗4.3.2 | Not reported | 44.78 ↗Effort: max · MaxIndependent result |
| AA-Briefcase ↗1.1 | Not reported | 1,506.86 ↗Effort: max · MaxIndependent result |
| GDPval-AA ↗2.1 | Not reported | 1,650.31 ↗Effort: max · MaxIndependent result |
| AutomationBench-AA ↗AA Index 4.3.2 collection | Not reported | 62.2% ↗Effort: max · MaxIndependent result |
| Terminal-Bench 4.0 · AA ↗ | Not reported | 41.92% ↗Effort: max · MaxHarness unreportedIndependent result |
| SciCode ↗AA Index 4.3.2 collection | Not reported | 59.03% ↗AA · Under reviewEffort: max · MaxIndependent result |
| Humanity’s Last Exam ↗AA Index 4.3.2 collection | Not reported | 42.26% ↗Effort: max · MaxIndependent result |
| GDP.pdf ↗AA Index 4.3.2 collection | Not reported | 11.2% ↗Effort: max · MaxIndependent result |
| CritPt ↗AA Index 4.3.2 collection | Not reported | 19.14% ↗AA · Under reviewEffort: max · MaxIndependent result |
| AA-Omniscience ↗AA Index 4.3.2 collection | Not reported | 14.3 ↗Effort: max · MaxIndependent result |
| AA-LCR ↗1.1 | Not reported | 79.67% ↗Effort: max · MaxIndependent result |
| LiveBench ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 60.74% ↗Effort: max · single run · equal mean of 7 suitesIndependent result |
| LiveBench Design ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 77.3% ↗Effort: max · single runIndependent result |
| LiveBench Research ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 73.2% ↗Effort: max · single runIndependent result |
| LiveBench Browser Use ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 0% ↗Effort: max · single runIndependent result |
| LiveBench Coding ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 82.8% ↗Effort: max · single runIndependent result |
| LiveBench Agentic ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 35.9% ↗Effort: max · single runIndependent result |
| LiveBench Data Analysis ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 70.2% ↗Effort: max · single runIndependent result |
| LiveBench Reasoning ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 85.8% ↗Effort: max · single runIndependent result |
Specifications | ||
| Context window | 131.1K | 1M |
| Max output | 118K | 943.7K |
| Parameters | 4B | — |
| Tool calling | Supported | Supported |
| Structured output | Supported | Supported |
| Image input | Supported | Not reported |
| Reasoning | Not reported | Supported |
| Prompt caching | Supported | Supported |
| Open weights | Supported | Supported |