Berget
Hybrid7 models hosted(3 own)ZDREU Infra
Avg Price
$1.33/M
Range
$0.013-$7.97
Reasoning
3
With AA Score
3
Hosting
7 self-hosted
Tier Breakdown
Standard
$7.97/M
1 models
$7.97 - $7.97
Budget
$0.443/M
1 models
$0.443 - $0.443
Micro
$0.185/M
5 models
$0.013 - $0.443
ZDR: Zero data retention: Berget states prompts/outputs are never stored and data is never used for training. Provider-stated (not independently verified). Note: for closed-weights models they host (e.g. GLM, Kimi), the no-retention guarantee is Berget's own operating policy, not the upstream lab's.
EU Sovereign: EU-only infrastructure on Berget's own hardware in Sweden; Swedish jurisdiction; no US entity involvement; GDPR/NIS2/Schrems II positioning. Provider-stated.
Hosted Models (click column to sort)
| Model | Owner | Tier | Host | Quant | Input $/M | Output $/M | Blended $/M▲ | Cost / IQ | SIT Score |
|---|---|---|---|---|---|---|---|---|---|
| multilingual e5 large | Berget | Micro | Self | FP16 | $0.033 | $0 | $0.013 | N/A | N/A |
| multilingual e5 large instruct | Berget | Micro | Self | FP16 | $0.033 | $0.022 | $0.027 | N/A | N/A |
| bge reranker v2 m3 | Berget | Micro | Self | FP16 | $0.111 | $0.111 | $0.111 | N/A | N/A |
| Mistral Small 3.2 24B Instruct 2506 | Mistralai | Micro | Self | BF16 | $0.332 | $0.332 | $0.332 | N/A | N/A |
| Google: Gemma 4 31BREASONING | Micro | Self | BF16 | $0.277 | $0.553 | $0.443 | $2.69 | N/A | |
| Z.ai: GLM 5.3 FlashREASONING | Z.ai | Budget | Self | FP8 | $0.277 | $0.553 | $0.443 | $0.518 | N/A |
| MoonshotAI: Kimi K3.1 ProREASONING | Moonshot AI | Standard | Self | MXFP4 | $3.32 | $11.07 | $7.97 | $5.85 | N/A |