Reports and data
AI reports
Read summaries of which AI models perform best across Nordic languages over time.
350+AI models tracked
2026-06-18Uppdaterad
Meta: Llama 3.1 8B InstructCurrent top signal
9.7/10Poäng
Weekly insight
Reports highlight winners, changes and notable movements in the model landscape.
Independent data
The reports are based on our own benchmark results and have no commercial provider ties.
For teams
Use the reports to support model choices for writing, research, support and internal tools.
Trend signal
Repeated measurements are more useful than a single marketing claim or launch announcement.
Dagens ranking
Oberoende. Inga affiliateavtal eller kommersiella bindningar till AI-leverantörer.
| # | Modell | Tier | Nordiskt | Instruktion | t/s | Poäng | Pris/1M |
|---|---|---|---|---|---|---|---|
| 1 | Meta: Llama 3.1 8B Instructmeta-llama | Budget | 10.0 | 10.0 | — | 9.7 | — |
| 2 | OpenAI: GPT-5.3 Chatopenai | Mid-range | 10.0 | 10.0 | — | 9.5 | — |
| 3 | Mistral Large 2407mistralai | Mid-range | 10.0 | 10.0 | — | 9.2 | — |
| 4 | Cohere: Command R+ (08-2024)cohere | Premium | 10.0 | 10.0 | — | 9.0 | — |
| 5 | AI21: Jamba Large 1.7ai21 | Mid-range | 10.0 | 10.0 | — | 8.8 | — |
| 6 | DeepSeek: DeepSeek V4 Flashdeepseek | Budget | 10.0 | 10.0 | — | 8.7 | — |
| 7 | Anthropic: Claude Opus 4.8 (Fast)anthropic | Premium | 10.0 | 10.0 | — | 8.5 | — |
| 8 | IBM: Granite 4.0 Microibm-granite | Budget | 10.0 | 8.0 | — | 8.5 | — |
Utforska mer
Över 350 AI-modeller utvärderas varje dag. De starkaste resultaten visas för nordiskt språkarbete.