Skip to content

v0.3.0

Choose a tag to compare

View all tags
@projectious projectious released this 18 May 15:33
· 50 commits to main since this release
v0.3.0
9fe4ec0

Capability radar — major UX addition.

New section on 04 Strategy: a 6-axis radar chart letting you see, at a glance, how any model compares against the reference across:

  • Coding — implement / debug / refactor
  • Reasoning & Architecture — design / hard logic / multi-step planning
  • Knowledge & Research — depth / synthesis / frontier awareness
  • Communication & Docs — write / explain / diagram for humans
  • Multimodal — see / hear / produce non-text
  • Agentic — model-side coherence + calibration (harness-independent)

Each model gets a 1–5 rating per axis against a frozen rubric (in src/dashboard-context.md). 19 frontier+fast models seeded; legacy / open-weight stay null until Sage back-fills in the next briefing.

Weighted composite + focus presets. One-click Focus dropdown boosts any one axis to weight 2.0 ('Coding-focused', 'Architecture-focused', etc.), recomputes the composite, and re-sorts the leaderboard. Default 'Balanced' weights all axes equally.

Provider-independent rendering. Pure SVG, no chart-library dependency. Build size 143 → 159 KB.

Background:

  • Architecture decision: DEC-20260518_1038-PolishedOak
  • Discussion: DISC-20260518_1024-SnowyCompass (resolved)
  • The Agentic axis is intentionally model-side only — harness contribution (loops, tools, memory, sandbox) is already scored on the 02 Harnesses tab, so the radar doesn't double-count. Effective autonomy = model Agentic × harness infrastructure (a 'stack composite' is v0.4.0 territory).

Live: https://projectious-work.github.io/ai-market-research/

Assets 4
Loading