| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| Perplexity Pro | Ctx: 128K · CiteScore: 9.1 · Papers: 200M+ | ✔️ (5 búsq./día) | ❌ |
| Gemini 2.5 Pro | Ctx: 1M · Docs: 10+ formatos · STEM: 92% | ❌ | ❌ |
| Consensus | Academic Acc: 89% · Sources: 1B+ | ✔️ Total | ❌ |
| Elicit AI | Auto-Synthesis: 95% · PDF Analysis | ✔️ (límite) | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| MANUS-Research | Ctx: 64K · PubMed Acc: 82% · ArXiv: 78% | ✔️ (HF) | ✔️ (HF) |
| SciSpace Copilot | PDF Q&A · Citation Extraction | ✔️ (10 PDF/mes) | ❌ |
| Haystack | RAG Framework · Doc Search: 88% | ✔️ (GitHub) | ✔️ (GitHub) |
| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| GPT-4.5 Turbo | HumanEval: 82.3% · Repo Analysis: 91% | ❌ | ❌ |
| DeepSeek-Coder v2 | HumanEval: 80.5% · MultiLang: 40+ | ✔️ Total | ✔️ (HF) |
| Grok-4 Heavy | LiveCodeBench: 79.4% · Debug Acc: 87% | ❌ | ❌ |
| CodeGemma 1.1 | Python Focused · MBPP: 84.1% | ✔️ (demo) | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| MANUS-Coder 7B | HumanEval: 76% · CodeClarity: 8.2/10 | ✔️ (HF) | ✔️ (HF) |
| WizardCoder-34B | HumanEval: 79.1% · Python: 92% | ✔️ (HF) | ✔️ (HF) |
| CodeLlama 3 70B | GSM8K-Python: 81.3% · SWE-bench: 68% | ✔️ (HF) | ✔️ (HF) |
| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| MidJourney v6.5 | Estética: 9.3/10 · Gen Speed: 2s | ❌ | ❌ |
| DALL·E 4 | Edit Real-Time · BLEU-Image: 41.2 | ✔️ (15 gen/día) | ❌ |
| Ideogram 2.5 | Text Legibility: 95% · Styles: 100+ | ✔️ (100 gen/mes) | ❌ |
| Playground v3 | Realism: 8.9/10 · Free Tier: 20 gen/día | ✔️ | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| MANUS-Diffusion | FID: 18.3 · CLIP: 32.1 | ✔️ (HF) | ✔️ (HF) |
| Stable Diffusion 3.5 | Custom LoRAs · Community Models | ✔️ (Local) | ✔️ (GitHub) |
| Fooocus | 1-Click Install · SDXL Optimized | ✔️ Total | ✔️ (GitHub) |
| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| Runway Gen-3 Alpha | 1080p · Motion Smoothness: 9.1/10 | ❌ | ❌ |
| Pika Labs 1.2 | Beta Abierta · Styles: 50+ | ✔️ | ❌ |
| Google Veo 3 | 120s Clips · Physics Accuracy: 94% | ❌ | ❌ |
| Kaiber 2.0 | Music Sync · 30s gratis/día | ✔️ | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| MANUS-Video | UCF101: 82% · Kinetics Acc: 76% | ✔️ (HF) | ✔️ (HF) |
| Stable Video | Apache 2.0 · 576x1024 Resolution | ✔️ (HF) | ✔️ (HF) |
| VideoCrafter 2 | 4K Support · GitHub Active | ✔️ Total | ✔️ (GitHub) |
| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| Claude 3.5 Opus | Ctx: 200K · MMLU: 89.3% · MATH: 84% · AGIEval: 45.1 | ✔️ (límite) | ❌ |
| Gemini 2.5 Pro | Ctx: 1M · MMLU: 88.5% · MATH: 81.2% · ARC: 91.7% | ✔️ (Gemini Pro) | ❌ |
| Grok-4 | Ctx: 128K · MMLU: 79% · MATH: 77.5% · HellaSwag: 89% | ✔️ (X Premium) | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| DeepSeek-R1 | Ctx: 128K · MMLU: 83.1% · MATH: 97.3% · GSM8K: 92.5% | ✔️ Total | ✔️ |
| Qwen 1.5 110B | Ctx: 128K · MMLU: 81.5% · MATH: 58.3% · C-Eval: 84% | 🌐 (demo) | ✔️ |
| Mixtral-8x22B | Ctx: 64K · MMLU: 81.4% · ARC: 85.2% · HellaSwag: 87.3% | 🌐 | ✔️ |
| Modelo Principal | Benchmarks | 🆓 Free Tier | 🔓 Open-Source |
|---|---|---|---|
| GPT-4.5 | SWE-bench: 82.1% · AgentBench: 8.7 · ToolCall: 96% · Ctx: 128K | ❌ | ❌ |
| Claude 3.5 | AgentBench: 8.9 · WebArena: 72.4% · Ctx: 200K · ToolCall: 94% | ✔️ (límite) | ❌ |
| Gemini 2.5 Pro | Ctx: 1M · AgentBench: 9.1 · SWE-bench: 78% · ToolCall: 97% | ❌ | ❌ |
| ⇩ Alternativas OS/Gratis ⇩ | |||
| DeepSeek-R1 | AutoGen: 85% · AgentBench: 8.2 · Ctx: 128K · SWE-bench: 75.6% | ✔️ Total | ✔️ |
| Kimi K2 | Ctx: 128K · AgentBench: 7.9 · ToolCall: 86% · WebArena: 68% | ✔️ 30 min/día | 🌐 (API) |
| MANUS-Agent | ToolCall: 84% · AgentBench: 7.7 · Ctx: 64K · WebArena: 65% | ✔️ | ✔️ |