Model comparison
Claude Fable 5.1 and Gemini 4 Argon compared on price, context window, capabilities and published benchmarks. Data comes from the OpenModels registry.
| Specification | Claude Fable 5.1 | Gemini 4 Argon |
|---|---|---|
| Developer | Anthropic | |
| Released | Sep 1, 2026 | Oct 1, 2026 |
| Context window | 1M tokens | 1.05M tokens |
| Input price / 1M | $10.00via Anthropic | $2.00via Google AI Studio |
| Output price / 1M | $50.00via Anthropic | $10.00via Google AI Studio |
| Providers | 1 | 2 |
| License | proprietary | proprietary |
| Input modalities | code, image, text | audio, code, image, text, video |
| Capabilities | chat, completion, function-calling, vision, code-generation, reasoning | chat, completion, function-calling, vision, audio, code-generation, reasoning |
| Reasoning | Yes | Yes |
| Tool calling | Yes | Yes |
Only benchmarks with a published score for at least one of the two models are shown. Published scores with their sources. Vendor-reported results use the vendor's own evaluation settings and are not directly comparable across vendors.
| Benchmark | Claude Fable 5.1 | Gemini 4 Argon |
|---|---|---|
| Artificial Analysis Intelligence Index | 53Artificial Analysis · Independent | 53Artificial Analysis · Independent |
| AutomationBench | 31.4%Anthropic · Vendor-reported | 51.3%Google (via Help Net Security) · Vendor-reported |
| Chartography | 88.4%Anthropic · Vendor-reported |
| — |
| CursorBench 4.0 | 51.8%Anthropic · Vendor-reported | — |
|---|
| CWE-bench | — | 68%Google (via Help Net Security) · Vendor-reported |
|---|
| DeepSWE 1.1 | — | 77.9%Google (via Help Net Security) · Vendor-reported |
|---|
| FrontierCode Main 1.1 | 50.3%Anthropic · Vendor-reported | — |
|---|
| GDPval-AA 2.1 | 1,735Anthropic · Vendor-reported | — |
|---|
| Humanity's Last Exam | 65.6%Anthropic · Vendor-reported | — |
|---|
| LVBench | — | 91.7%Google (via Help Net Security) · Vendor-reported |
|---|
| OSWorld 2.1 | 80.7%Anthropic · Vendor-reported | — |
|---|
| Terminal-Bench 4.0 | 55.8%Anthropic · Vendor-reported | — |
|---|
| Terminal-Bench-Science 0.1 | 52.6%Anthropic · Vendor-reported | — |
|---|