Side by side
See the difference.
Scores, capabilities and costs for the models you’re considering.
Scroll to compare models →
| 3 models | fal-aiSa2VA 8B ImageRemove × | googleGoogle: Gemini 4 ArgonRemove × | anthropicAnthropic: Claude Sonnet 5.5Remove × |
|---|---|---|---|
| What’s different | Sa2VA is an MLLM capable of question answering, visual prompt understanding, and dense object segmentation at both image and video levels | Long decode can use the unified 1M-token budget for output; reads video, audio and PDFs, with implicit-only caching. | Tuned for quick iteration on scoped fixes and documents; Anthropic favors Opus 5.5 for open-ended work needing sustained judgment. |
| Date | Jan 13, 2025Listed | Sep 30, 2026Released | Sep 28, 2026Released |
API prices USD · provider configuration shown below | |||
| Listed rate | $0.04/ imagefal · all routes ↗ | $1.00 in / $5.00 out/ 1M tokensGoogle AI Studio · via OpenRouter · all routes ↗google-ai-studio/flex · tier=flex | $2.00 in / $10.00 out/ 1M tokensAmazon Bedrock · via OpenRouter · all routes ↗amazon-bedrock |
| Provider endpoints | 1 endpoint → | 8 endpoints → | 11 endpoints → |
Benchmarks Each result retains its evaluation configuration | |||
| AA Intelligence ↗4.3.2 | Not reported | 52.56 ↗Effort: high · HighIndependent result | 56 ↗Effort: max · Max, Default FallbackIndependent result |
| AA-Briefcase ↗1.1 | Not reported | 1,485.76 ↗Effort: high · HighIndependent result | 1,816.92 ↗Effort: max · Max, Default FallbackIndependent result |
| GDPval-AA ↗2.1 | Not reported | 1,623.11 ↗Effort: high · HighIndependent result | 1,837.82 ↗Effort: max · Max, Default FallbackIndependent result |
| AutomationBench-AA ↗AA Index 4.3.2 collection | Not reported | 77.51% ↗Effort: high · HighIndependent result | 71.77% ↗Effort: max · Max, Default FallbackIndependent result |
| Terminal-Bench 4.0 · AA ↗ | Not reported | 57.07% ↗Effort: high · HighHarness unreportedIndependent result | 63.64% ↗Effort: max · Max, Default FallbackHarness unreportedIndependent result |
| SciCode ↗AA Index 4.3.2 collection | Not reported | 61.81% ↗AA · Under reviewEffort: high · HighIndependent result | 61% ↗AA · Under reviewEffort: max · Max, Default FallbackIndependent result |
| Humanity’s Last Exam ↗AA Index 4.3.2 collection | Not reported | 57.09% ↗Effort: high · HighIndependent result | 54.96% ↗Effort: max · Max, Default FallbackIndependent result |
| GDP.pdf ↗AA Index 4.3.2 collection | Not reported | 21.8% ↗Effort: high · HighIndependent result | 25.8% ↗Effort: max · Max, Default FallbackIndependent result |
| CritPt ↗AA Index 4.3.2 collection | Not reported | 27.14% ↗AA · Under reviewEffort: high · HighIndependent result | 31.43% ↗AA · Under reviewEffort: max · Max, Default FallbackIndependent result |
| AA-Omniscience ↗AA Index 4.3.2 collection | Not reported | 42.35 ↗Effort: high · HighIndependent result | 32.3 ↗Effort: max · Max, Default FallbackIndependent result |
| AA-LCR ↗1.1 | Not reported | 79.67% ↗Effort: high · HighIndependent result | 82.67% ↗Effort: max · Max, Default FallbackIndependent result |
| LiveBench ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 84.1% ↗Effort: high · single run · equal mean of 7 suitesIndependent result | 82.46% ↗Effort: max · single run · equal mean of 7 suitesIndependent result |
| LiveBench Design ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 73.7% ↗Effort: high · single runIndependent result | 85% ↗Effort: max · single runIndependent result |
| LiveBench Research ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 83.1% ↗Effort: high · single runIndependent result | 83.6% ↗Effort: max · single runIndependent result |
| LiveBench Browser Use ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 92% ↗Effort: high · single runIndependent result | 84% ↗Effort: max · single runIndependent result |
| LiveBench Coding ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 89.7% ↗Effort: high · single runIndependent result | 94.7% ↗Effort: max · single runIndependent result |
| LiveBench Agentic ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 69.9% ↗Effort: high · single runIndependent result | 78.8% ↗Effort: max · single runIndependent result |
| LiveBench Data Analysis ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 85.2% ↗Effort: high · single runIndependent result | 59.5% ↗Effort: max · single runIndependent result |
| LiveBench Reasoning ↗2.0 · 2026-10-14 (dataset 2026-09-02) | Not reported | 95.1% ↗Effort: high · single runIndependent result | 91.6% ↗Effort: max · single runIndependent result |
Specifications | |||
| Context window | — | 1M | 1M |
| Max output | — | 943.7K | 128K |
| Parameters | — | — | — |
| Tool calling | Not reported | Supported | Supported |
| Structured output | Not reported | Supported | Supported |
| Image input | Not reported | Supported | Supported |
| Audio input | Not reported | Supported | Not reported |
| Video input | Not reported | Supported | Not reported |
| Reasoning | Not reported | Supported | Supported |
| Prompt caching | Not reported | Supported | Supported |