Example

Example decision brief

A worked example of the evidence pack the /briefs/build workspace produces. Two verified models from the long-context-analysis demo, rendered with the same helper that powers the live builder and the export endpoint. Evidence, not a recommendation.

Preview

What the brief contains

Server-rendered from buildDecisionBrief(). Open the export links below to see the full Markdown or JSON payload.

Example brief — long-context analysis

Decision brief — Long-context analysis

Generated at: 2026-05-24 · Use case: Long-context analysis

Models
2
Evidence rows
16
Data gaps
2
Sources
3

Show selected models

Gemini 2.5 Pro — Google · lifecycle: active
Claude Haiku 4.5 — Anthropic · lifecycle: active

Open builder →Export Markdown Export JSON

Preview generated from current catalogue data. Briefs are evidence, not recommendations.

Evidence

Verified fields, side by side

The same verified-data layer the comparison builder reads. Unknown values render an em-dash.

Example evidence — long-context candidates

Model	Lifecycle	Context	Modalities	Hosted	Sources	Freshness
Gemini 2.5 Pro	active	1,048,576	text-in, image-in, audio-in, video-in, text-out	—	3	Fresh
Claude Haiku 4.5	active	200,000	text-in, image-in, text-out	—	2	Fresh

Preview generated from current catalogue data. Every value is rendered straight from the verified data layer; unknown values stay null.

Honest gaps

Data gaps in this brief (2)

Fields the catalogue records as unverified. The brief surfaces them explicitly rather than inventing values.

gemini-2-5-pro · release date — Vendor docs not yet retrieved or value not published.
claude-haiku-4-5 · release date — Vendor docs not yet retrieved or value not published.

Sources

Source trail (3)

Every primary-source citation referenced by the brief. Open each link to confirm the value on the vendor page.

ai.google.dev/gemini-api
Google AI — Gemini 2.5 Pro model docs
official-vendor-docs · retrieved 2026-05-20
ai.google.dev/pricing
Google AI — Gemini pricing
official-vendor-pricing · retrieved 2026-05-20
platform.claude.com/docs
Anthropic — Models overview
official-vendor-docs · retrieved 2026-05-20

Next external tests (still your job)

Run task-specific prompt tests against the candidate models in your own environment.
Verify request latency from your target deployment region.
Check rate limits in your provider account against your expected load.
Validate per-token cost against the vendor's current pricing page (references in this brief are not live quotes).
Confirm compliance / security requirements against your organisation's controls — verification status here is not a certification.

Build your own brief

The example above is one pre-built path. Start from any use case or pick your own model set:

/briefs/build — open the builder, pick up to four models, optionally choose a use case + field set, and click Build.
/api/briefs/decision?models=gemini-2-5-pro%2Cclaude-haiku-4-5&useCase=long-context-analysis&format=markdown — Markdown export endpoint with the same query shape.
/api/briefs/decision?models=gemini-2-5-pro%2Cclaude-haiku-4-5&useCase=long-context-analysis&format=json — JSON export endpoint.
/demos — three guided demos that walk the full five-step workflow.
/docs/decision-briefs — what a brief is and what it deliberately does not assert.