AIDeskPro GenAI Engines
The best LLMs available, selected and tested for you
Our tagline is “The best GenAI engines in one Tool”. But which models are available in AIDeskPro? What are the differences between them?
Saying which is the best LLM is impossible: we have tested them all and each has its strengths and characteristics. We provide objective data from public benchmarks and leave the choice of the most suitable engine for your needs up to you.
Models marked with ✦ are already available in AIDeskPro.
Agent Leaderboard — September 2026
Dynamic ranking based on real sessions in Agent Mode. Measures how well each model orchestrates tools, completes autonomous tasks and responds to user corrections.
✦ Disponibile in AIDeskPro
✦ AIDeskPro
2
Claude Opus 5.5 (High)
Anthropic
11.84%
✦ AIDeskPro
1
Claude Fable 5.1 (Max)
Anthropic
14.06%
✦ AIDeskPro
3
GPT 6 Astra (Max)
OpenAI
10.36%
| # | Modello | Net Impr. | Confirmed | Praise | Steer. | Recovery | Halluc. ↓ |
|---|---|---|---|---|---|---|---|
| 1 | ✦ Claude Fable 5.1 (Max) Anthropic | ▲ 14.06% | 17.37% | 32.43% | 8.40% | 11.78% | 0.34% |
| 2 | ✦ Claude Opus 5.5 (High) Anthropic | ▲ 11.84% | 16.25% | 19.28% | 13.57% | 9.81% | 0.30% |
| 3 | ✦ GPT 6 Astra (Max) OpenAI | ▲ 10.36% | 12.68% | 33.92% | ▼ 1.39% | 6.26% | 0.35% |
| 4 | ✦ Claude Opus 5 (Max) Anthropic | ▲ 9.54% | 10.18% | 16.64% | 7.91% | 12.62% | 0.34% |
| 5 | ✦ Claude Opus 5 (High) Anthropic | ▲ 9.38% | 6.69% | 17.66% | 10.68% | 11.54% | 0.31% |
| 6 | ✦ GPT 6 Sol (Max) OpenAI | ▲ 8.80% | 10.84% | 10.51% | 14.54% | 7.75% | 0.35% |
| 7 | ✦ Claude Fable 5 (High) Anthropic | ▲ 7.99% | 3.10% | 16.58% | 11.43% | 8.49% | 0.35% |
| 8 | ✦ Claude Opus 4.8 (High) Anthropic | ▲ 6.87% | 3.85% | 15.19% | 9.59% | 5.75% | 0.05% |
| 9 | ✦ GPT 5.6 Sol (xHigh) OpenAI | ▲ 6.26% | 3.09% | 18.82% | 6.54% | 2.51% | 0.35% |
| 10 | ✦ Claude Sonnet 5 (High) Anthropic | ▲ 4.80% | 0.79% | 9.11% | 7.49% | 6.44% | 0.18% |
| 11 | ✦ GPT 5.5 (xHigh) OpenAI | ▲ 4.24% | ▼ 2.61% | 7.82% | 6.86% | 8.80% | 0.35% |
| 12 | Hy4 preview Tencent | ▲ 4.10% | 7.53% | 7.45% | ▼ 0.70% | 6.22% | 0.02% |
| 13 | ✦ Kimi K3 (Max) Moonshot | ▲ 4.03% | 9.18% | 7.35% | ▼ 0.54% | 3.83% | 0.35% |
| 14 | ✦ Grok 4.7 (xHigh) SpaceXAI | ▲ 4.02% | 8.99% | 4.42% | 0.46% | 5.88% | 0.35% |
| 15 | ✦ Deepseek V4.1 Flash (Max) DeepSeek | ▲ 3.86% | 12.08% | 2.90% | ▼ 3.29% | 7.39% | 0.22% |
| 16 | ✦ GLM 5.2 (Max) Z.ai | ▲ 3.84% | 3.78% | 9.84% | 4.18% | 1.08% | 0.35% |
| 17 | Muse Spark 1.3 (Max) Meta | ▲ 3.83% | 8.35% | 3.67% | 1.58% | 5.20% | 0.35% |
| 18 | ✦ Gemini 3.8 Flash (High) Google | ▲ 3.75% | 9.30% | 5.06% | 4.26% | ▼ 0.06% | 0.21% |
| 19 | ✦ GLM 5.3 (Max) Z.ai | ▲ 2.78% | 5.99% | 5.69% | 2.81% | ▼ 0.93% | 0.35% |
| 20 | ✦ Qwen3.8 Max Alibaba | ▲ 2.46% | 5.19% | 3.67% | 2.03% | 2.52% | 1.11% |
| 21 | ✦ DeepSeek V4 Pro (High) (0813) DeepSeek | ▲ 1.74% | ▼ 1.28% | 3.11% | 1.76% | 4.77% | 0.35% |
| 22 | ✦ GPT 5.5 OpenAI | ▲ 1.68% | ▼ 4.47% | 4.90% | 2.26% | 5.35% | 0.35% |
| 23 | ✦ GPT 6 Luna (Max) OpenAI | ▲ 1.59% | 4.63% | ▼ 2.77% | 1.51% | 4.24% | 0.35% |
| 24 | ✦ Grok 4.5 SpaceXAI | ▲ 1.25% | ▼ 1.07% | ▼ 2.71% | 3.86% | 5.84% | 0.35% |
| 25 | ✦ Grok 4.6 (xHigh) SpaceXAI | ▲ 1.13% | ▼ 3.32% | ▼ 0.24% | 3.26% | 5.61% | 0.35% |
| 26 | ✦ GPT 5.6 Terra (xHigh) OpenAI | ▲ 0.64% | ▼ 6.36% | 2.75% | 4.11% | 2.34% | 0.35% |
| 27 | ✦ GPT 5.4 (High) OpenAI | ▲ 0.31% | ▼ 3.50% | ▼ 0.73% | 2.43% | 3.00% | 0.35% |
| 28 | ✦ GLM 5.3 Flash Z.ai | ▼ 0.08% | 6.64% | ▼ 0.44% | ▼ 1.06% | ▼ 5.88% | 0.35% |
| 29 | ✦ GPT 5.6 Luna (xHigh) OpenAI | ▼ 0.86% | ▼ 4.60% | ▼ 4.18% | 1.92% | 2.20% | 0.35% |
| 30 | ✦ Qwen3.8 Flash Next Alibaba | ▼ 0.88% | 5.52% | ▼ 2.87% | ▼ 3.42% | ▼ 2.77% | 0.90% |
| 31 | ✦ DeepSeek V4 Pro DeepSeek | ▼ 1.23% | ▼ 3.71% | ▼ 1.59% | ▼ 2.82% | 2.16% | 0.18% |
| 32 | ✦ Gemini 3.7 Flash (High) Google | ▼ 1.31% | 0.27% | ▼ 3.28% | ▼ 0.01% | ▼ 3.45% | 0.06% |
| 33 | ✦ Qwen 3.8 27B Alibaba | ▼ 1.54% | 2.44% | ▼ 2.63% | ▼ 0.97% | ▼ 6.28% | 0.28% |
| 34 | Muse Spark 1.2 (xHigh) Meta | ▼ 3.24% | ▼ 5.38% | ▼ 11.66% | ▼ 5.40% | 5.90% | 0.32% |
| 35 | Muse Spark 1.1 Meta | ▼ 4.07% | ▼ 1.98% | ▼ 11.10% | ▼ 5.62% | ▼ 1.97% | 0.33% |
| 36 | ✦ Qwen3.7 Max Alibaba | ▼ 4.59% | ▼ 9.18% | ▼ 8.35% | ▼ 3.46% | ▼ 1.73% | 0.22% |
| 37 | Hy3 Tencent | ▼ 5.75% | ▼ 11.30% | ▼ 7.34% | ▼ 1.35% | ▼ 5.91% | 2.85% |
| 38 | Mimo V2.5 Pro Xiaomi | ▼ 6.61% | ▼ 11.12% | ▼ 10.56% | ▼ 4.52% | ▼ 5.90% | 0.92% |
| 39 | Minimax M3 MiniMax | ▼ 6.81% | ▼ 14.28% | ▼ 13.18% | ▼ 6.99% | 0.51% | 0.09% |
| 40 | ✦ Gemini 3.6 Flash (High) Google | ▼ 6.93% | ▼ 7.58% | ▼ 8.52% | ▼ 3.98% | ▼ 14.90% | 0.31% |
| 41 | ✦ Gemini 3.1 Pro Preview Google | ▼ 7.22% | ▼ 7.96% | ▼ 0.70% | ▼ 3.98% | ▼ 23.03% | 0.43% |
| 42 | ✦ Qwen3.7 Plus Alibaba | ▼ 7.23% | ▼ 9.47% | ▼ 12.51% | ▼ 7.47% | ▼ 5.68% | 1.02% |
| 43 | Inkling Small Thinking Machines | ▼ 9.60% | ▼ 21.26% | ▼ 20.52% | ▼ 11.51% | 5.72% | 0.42% |
| 44 | Inkling Thinking Machines | ▼ 11.00% | ▼ 21.89% | ▼ 22.04% | ▼ 11.79% | 0.73% | 0.03% |
| 45 | Mistral Medium 3.5 Mistral | ▼ 11.95% | ▼ 16.36% | ▼ 15.47% | ▼ 7.18% | ▼ 18.26% | 2.48% |
| 46 | Solar Pro 4 Upstage | ▼ 16.25% | ▼ 23.15% | ▼ 20.12% | ▼ 10.33% | ▼ 25.73% | 1.89% |
Cloud Providers
AIDeskPro as a service always resides in Europe. All default actions in the Tool use LLMs with EU data residency and zero data retention. Within the Tool, some models with global endpoints are also available, labelled “No EU”: you can choose them knowingly for non-critical operations. Here are the cloud partners that allow us to offer the best models on the market.
Google Cloud & Model Garden
As a Google Premier Partner we guarantee data hosted in Italy (Milan) or in Europe, access to Gemini in all versions and to Anthropic models via Model Garden, with zero data retention.
Scopri di più →Anthropic
Claude models (Sonnet, Opus, Haiku, Fable) are available in AIDeskPro via Google Model Garden with EU residency guarantee and no use of data for training for paid versions.
Scopri di più →OpenAI
GPT models are available through an Enterprise agreement with Zero Data Retention and European Data Residency. These guarantees do not apply to consumer ChatGPT accounts.
Scopri di più →AWS — Amazon Web Services
AIDeskPro can be deployed on AWS in European Regions (Frankfurt, Ireland, Stockholm) with Amazon Bedrock integration, GDPR compliance and enterprise certifications.
Scopri di più →Scaleway
100% European cloud provider with data centres in Paris, Amsterdam and Warsaw. Ideal for companies with European digital sovereignty requirements, powered by renewable energy.
Scopri di più →Nebius AI
Next-generation AI cloud platform with H100/H200 GPUs, data centres in Europe (Helsinki, Amsterdam). Ideal for open source models (Llama, Mistral, Qwen…) in a private environment.
Scopri di più →Control and transparency on privacy
AIDeskPro gives you full visibility into where your data is processed. For each model, you know exactly which endpoint is used:
EU DATA RESIDENCY
AIDeskPro as a service always runs in Europe. All default actions use LLMs with European endpoints and guaranteed data residency. Within the Tool, models with global endpoints are also available: they are explicitly labelled “No EU” and are always an active user choice - never the default.
ZERO DATA RETENTION
No provider uses your data, documents or conversations to train their models. Contractual guarantee with all partners.
ENTERPRISE AGREEMENTS
All providers we work with have Enterprise agreements with Open Gate, with specific clauses on privacy, security and GDPR compliance.