A compact research index for important AI model releases across language, image, video, audio, multimodal systems, and world models. Follow a model to its evidence—or a benchmark to every comparable result.
Latest additions
Updates
Release records are the timeline. New reports and evaluation evidence can update an existing model without creating a duplicate.
Browse by model
Models
The record follows the exact model version. Related models connect family history, modality, market position, and shared evaluations.
Daily-use Opus generation focused on judgment, efficiency, coding, scientific research, and professional work.
languagereasoningcodingagentmultimodal
Released
Jul 24, 2026
Status
production
Context
Not disclosed
Access
api, app
Input
text, image
Output
text
Research note
Anthropic positions Opus 5 near Fable-level intelligence at a lower operating cost. Its release emphasizes artifact quality, verification behavior, and cost per completed task rather than one aggregate intelligence score.
Evidence gap
The release includes many partner and internal evaluations, but comparable third-party evidence is still early. Preserve effort, harness, attempts, and cost with every result.
Evaluation evidence
0 measurements
No numeric evidence published into the research index yet.
Related models
family · modality · shared task
Browse by evidence
Benchmarks
All reported models are visible, but only materially aligned protocols share a ranking group. Source verification and protocol comparability are tracked separately.
terminal bench 2 1 provider maxreported only · 5 models