How to read this comparison
“Primary” means the price comes from the provider's pricing page. “Secondary” means the pricing page could not be read automatically, so the figure comes from several reports and should be checked with the provider before you buy. Prices are in the provider's currency, without guarantee.
Plans
| Free | Free | |
|---|
| Go | about $8 / month | Reportedly this plan may include ads. |
|---|
| Plus | $20 / month | |
|---|
| Pro | from $100 / month | Tiered by usage. |
|---|
Price notes: Secondary source – Prices come from secondary sources because OpenAI's pricing page blocks automated access.
Strengths
- Largest ecosystem of add-ons and integrations
- Current model family GPT-6 (incl. GPT-6 Astra)
- Broad API with batch and tool features
Weaknesses
- High tiers for heavy usage
- Check data protection for personal data in the EU first
Well suited for
- All-round assistant
- Research and analysis
- Image and voice features
- Current models
- GPT-6 Astra · GPT-6.1 Sol (laut Presseberichten) · GPT-5.6 Luna / Terra / SolBest HLE result: 53.6 % (GPT-6 Astra)
- API: Yes
- Business and Enterprise plans exist for companies. Check the settings for using data for training.
Sources: OpenAI: ChatGPT-Preise (Anbieterseite, Abruf blockiert, bitte dort prüfen) · OpenAI-Modelle im HLE-Ranking
Plans
| Free | $0 | Sonnet and Haiku models. |
|---|
| Pro | $20 / month ($17 billed annually) | |
|---|
| Max | from $100 / month | Options with 5x and 20x the usage of Pro. |
|---|
| Team | $25 per seat / month ($20 billed annually) | Premium seats: $125 ($100 annually). |
|---|
| Enterprise | $20 per seat (annual) plus usage at API rates | |
|---|
Price notes: Primary source – API per 1M tokens (input / output): Fable 5.1 $10 / $50, Opus 5.5 $4 / $20, Sonnet 5.5 $2 / $10, Haiku 5.5 from $0.10 / $0.50.
Strengths
- Four model tiers: Fable 5.1, Opus 5.5, Sonnet 5.5, Haiku 5.5
- API prices published per model
Weaknesses
- Fable 5.1 is the most expensive API model in this comparison ($10 / $50 per M tokens)
- Usage limits in the plans cap heavy use
Well suited for
- Long texts and documents
- Coding
- Analysis and writing
- Current models
- Fable 5.1 · Opus 5.5 · Sonnet 5.5 · Haiku 5.5Best HLE result: 55.4 % (Opus 5.5)
- API: Yes
- Business plans (Team, Enterprise) come with their own data commitments; check the provider's terms.
Sources: Anthropic: Claude-Preise
Plans
| Google AI Plus | €4.99 / month | 400 GB storage, 2x usage limits. |
|---|
| Google AI Pro | €21.99 / month | 5 TB storage, 4x usage limits. |
|---|
| Google AI Ultra | €99.99 or €219.99 / month | Tiers with 5x and 20x limits, from 20 TB storage, Deep Think. |
|---|
Price notes: Primary source
Strengths
- Tightly integrated with Gmail, Docs and other Google services
- Plans include storage (400 GB to 20+ TB)
Weaknesses
- Ultra plans are expensive
- Current model names change quickly
Well suited for
- Google Workspace users
- Multimodal tasks
- Research with Google integration
- Current models
- Gemini 3.1 Pro · Gemini 3.8 Flash · Gemini 3.7 FlashBest HLE result: 55.6 % (Gemini 4 Argon)
- API: Yes
- Available to companies via Google Workspace and Google Cloud; processing there follows the cloud terms.
Sources: Google: KI-Tarife (Österreich)
Plans
| Microsoft 365 Premium | $19.99 / month | Replaces the former Copilot Pro for individuals. |
|---|
| Microsoft 365 Copilot Business | $18 / user / month (promotional, regular $21) | For organisations up to 300 users, promotion until 31 Dec 2026. |
|---|
| Microsoft 365 Copilot | $30 / user / month (annual) | Add-on to an existing Microsoft 365 licence. |
|---|
Price notes: Secondary source – Prices according to secondary sources (Oct 2026); Microsoft 365 base prices rose on 1 July 2026, so check total cost with your reseller.
Strengths
- Directly inside Word, Excel, PowerPoint, Teams and Outlook
- EU Data Boundary for Microsoft 365 customers
- Access to company data in your own tenant
Weaknesses
- Extra licence cost on top of Microsoft 365
- Value depends heavily on well-maintained data
Well suited for
- Microsoft 365 organisations
- Summarising meetings
- Automating documents and spreadsheets
- Current models
- Wechselnde Modelle verschiedener Anbieter (nicht einzeln wählbar in allen Apps)
- API: Yes
- Microsoft 365 applies enterprise data protection and the EU Data Boundary; tenant data stays in the tenant.
Sources: Microsoft: Copilot-Pläne · Microsoft: EU Data Boundary
Plans
| Free | $0 | 2,000 inline suggestions per month, limited chat and agent use. |
|---|
| Pro | $10 / month | $15 AI credits per month, unlimited inline suggestions. |
|---|
| Pro+ | $39 / month | $70 AI credits per month. |
|---|
| Max | $100 / month | $200 AI credits per month. |
|---|
| Business / Enterprise | per user / on request | Business with management and IP indemnity, Enterprise with codebase indexing. |
|---|
Price notes: Primary source
Strengths
- Inside editors, CLI and GitHub
- All plans support agent mode, CLI and GitHub integration
- Models from several providers (incl. Claude, GPT, Gemini) selectable
Weaknesses
- Specialised in programming
- Billing via AI credits instead of unlimited use
Well suited for
- Software development
- Code reviews
- Tests and documentation
- Current models
- Mehrere Anbieter wählbar
- API: Limited
- Business and Enterprise with organisation management and IP indemnity according to the provider.
Sources: GitHub: Copilot-Tarife
Plans
| Free | $0 | |
|---|
| Pro | $20 / month | |
|---|
| Max | $200 / month | Highest usage limits, Comet browser and more. |
|---|
Price notes: Secondary source – Sonar API according to a secondary source: Sonar $1 / $1, Sonar Pro $3 / $15 per M tokens, plus search fees.
Strengths
- Answers with source citations
- Live web search as the core of the product
- Sonar API for your own applications
Weaknesses
- Less focused on creative writing and long conversations
- Spot-check the cited sources
Well suited for
- Research with citations
- Factual questions
- Market monitoring
- Current models
- Sonar, Sonar Pro (API)
- API: Yes
- No statements I can verify here: check the provider's privacy terms.
Sources: Perplexity: Preise (Anbieterseite, Abruf blockiert, bitte dort prüfen)
Plans
| Free | $0 | Limited messages, web search and image generation. |
|---|
| Pro | $14.99 / month (students $5.99) | Includes monthly API credit. |
|---|
| Team | $24.99 / user / month (min. $50) | 30 GB storage per user. |
|---|
| Enterprise | on request | SSO, audit logs, custom models and agents. |
|---|
Price notes: Primary source – API according to the pricing page: Mistral Large $0.50 input / $1.50 output per M tokens; batch halves the price.
Strengths
- European provider
- Several models also with open weights (Apache 2.0)
- Low-cost Pro plan ($14.99)
Weaknesses
- Not part of the HLE ranking, so no direct performance comparison
- Smaller ecosystem than the big players
Well suited for
- GDPR-conscious users
- Multilingual text
- Self-hosting open models
- Current models
- Mistral Large 3 (offene Gewichte) · Mistral Medium 3.5 · Mistral Small 4
- API: Yes
- French provider; an obvious candidate for EU use. Check the data processing agreement case by case.
Sources: Mistral: Preise
Plans
| SuperGrok | about $30 / month | According to secondary sources; a higher tier with Heavy access is much more expensive. |
|---|
Price notes: Secondary source
Strengths
- Closely tied to the X platform
- Current flagship Grok 4.6 (Aug 2026)
Weaknesses
- Prices only backed by secondary sources
Well suited for
- Current topics on X
- Experimental use
- Current models
- Grok 4.6 · Grok 4.5Best HLE result: 39.7 % (Grok 4.6)
- API: Yes
- No verified statements: check xAI's privacy terms before use.
Sources: xAI: Grok · HLE-Ranking
Plans
| API deepseek-flash | $0.15–0.30 input / $0.60–1.20 output per M tokens | Time-dependent prices: half outside peak hours. |
|---|
| API deepseek-v4-pro | $0.66–1.32 input / $1.98–3.96 output per M tokens | |
|---|
Price notes: Primary source
Strengths
- Open weights under the MIT licence
- 1M tokens of context for both current models
- Very low API prices
Weaknesses
- Data protection when using the provider API
- In the HLE ranking (see statistics) DeepSeek 4 Pro trails the top US models
Well suited for
- Low-cost API use
- Self-hosted models
- Coding tasks and agents
- Current models
- DeepSeek V4.1 Flash · DeepSeek V4 ProBest HLE result: 32.4 % (DeepSeek 4 Pro)
- API: Yes
- With the provider's API, data is processed by the provider in China; do not use it for personal data in the EU without review. The open weights (MIT) can be self-hosted.
Sources: DeepSeek: API-Preise · DeepSeek auf Hugging Face