model evaluation

Everything tagged model evaluation.

Share your story Subscribe

Featured

News

Property Finder rebuilds AI benchmark, reports 86% lower inference cost

DP News TeamOct 28, 2026

News

Coding-agent harnesses can triple costs without clear accuracy gains, study finds

Oct 23, 2026

News

Anthropic pitches lawmakers on draft frontier AI safety bill

Oct 16, 2026

Latest

News
Anthropic opens Singapore office and signs government AI MOU
Anthropic opened its fifth Asia-Pacific office in Singapore and signed an MOU with the country's digital ministry on AI safety, evaluation, adoption…
DP News Team · 2w ago
News
Apple study finds LLM translation grading unreliable in most settings
Apple researchers found wide swings in how LLMs and people judged translation quality, with agreement unreliable in most settings tested.
DP News Team · 2w ago
News
CoreWeave launches Forge to link the production AI workflow
CoreWeave’s Forge brings serving, observability, data curation, post-training and evaluation into one environment for teams improving production…
DP News Team · 3w ago
News
TII launches Falcon-Emirati-7B for Emirati Arabic
Technology Innovation Institute launched Falcon-Emirati-7B, an adaptation of Falcon-H1-Arabic for Emirati speech and cultural context, without…
DP News Team · 3w ago
News
Apple study finds LLM conditioning gains can reduce fluency
An Apple-published study found that methods for conditioning language models can improve control while degrading fluency, with activation steering…
DP News Team · 4w ago
News
Anthropic and Accenture plan embedded AI safety team
Faculty is set to evaluate models from inside Anthropic with employee-like access, while access and reporting standards remain unsettled.
DP News Team · Sep 19
News
Google confirms Gemini accessed three companies during a safety test
Google confirmed that Gemini reached systems at three real companies during a May cybersecurity evaluation after failures in test isolation and…
DP News Team · Sep 18
News
OpenAI and Hugging Face disclose models escaped a cyber-eval and breached production systems
OpenAI and Hugging Face said two OpenAI models escaped a sandboxed cyber-capability test and breached Hugging Face's production systems using a real…
DP News Team · Jul 22
News
OpenAI introduces Deployment Simulation to predict model behavior before launch
The method replays about 1.3 million de-identified ChatGPT conversations against a candidate model, and caught a reward-hacking behavior OpenAI calls…
Dmytro Spodarets · Jun 17

Featured

Property Finder rebuilds AI benchmark, reports 86% lower inference cost

Coding-agent harnesses can triple costs without clear accuracy gains, study finds

Anthropic pitches lawmakers on draft frontier AI safety bill

Latest

Anthropic opens Singapore office and signs government AI MOU

Apple study finds LLM translation grading unreliable in most settings

CoreWeave launches Forge to link the production AI workflow

TII launches Falcon-Emirati-7B for Emirati Arabic

Apple study finds LLM conditioning gains can reduce fluency

Anthropic and Accenture plan embedded AI safety team

Google confirms Gemini accessed three companies during a safety test

OpenAI and Hugging Face disclose models escaped a cyber-eval and breached production systems

OpenAI introduces Deployment Simulation to predict model behavior before launch